Treat that as a workforce strategy because that’s what it is. The distributors, wholesalers and merchants that frame AI this way are already pulling ahead. The ones still running AI primarily as an IT project risk losing twice: first the people, then the knowledge they take with them.

Walk into almost any trade counter or branch in the UK and find the person everyone relies on. In a builders’ merchant, electrical wholesaler or industrial distributor, it is often someone in their 50s. They know which fittings actually cross-reference, which customer pays in 30 days and which one needs chasing, and which substitute to offer when a product is out of stock.

Almost none of it is written down.

I ran branch and call center operations at Grainger for the better part of a decade, and those were the people who kept the place standing. Last year, I watched a regional distributor lose three of them in a single quarter to retirement. The replacements were capable. They just didn’t know what the people who left knew. Service scores slipped, returns crept up, and the branch manager spent his days answering questions that used to answer themselves.

That’s the real AI story in our industry right now, and it has almost nothing to do with chatbots.

The Clock You’re Actually Racing

Here’s the uncomfortable part. The retirement wave may be the most important workforce issue on your five-year horizon, and many distribution leaders are managing it primarily as a hiring problem when it is also a knowledge problem. You can’t recruit your way out of a demographic cliff.

Look at the numbers in the trades that feed our sector. In UK construction and building materials, 35% of the workforce is over 50 and only 20% is under 30, with the average bricklayer now 52, according to 2025 data cited by ecomerchant. By 2035, more than a third of today’s workforce will reach retirement age, and roughly 750,000 workers are due to leave by 2036. Over the same period, the sector needs 251,500 additional workers by 2028 just to meet demand, against more than 140,000 unfilled vacancies as of December 2025.

The math doesn’t balance. The people aren’t there to hire.

This runs straight through the broader workforce supporting distributors, wholesalers and merchants. A March 2025 study by Flip and Workplace Intelligence, covering 500 UK frontline managers and employees in manufacturing and retail, found that 57% of the most experienced frontline workers are within five years of retirement. Sixty-eight percent of their managers fear vital expertise will be lost when those people go, and 78% aren’t confident their company is ready for the skills gap that follows.

The Organization for Economic Cooperation and Development made the broader demographic challenge clear in its 2025 Employment Outlook: Developed economies increasingly face labor scarcity as populations age and the ratio of older people to working-age populations rises.

Now set AI against that backdrop.

The Signal in the Data

Read that table from top to bottom and the strategy begins to write itself. The people who hold your operating knowledge are leaving. Distributors, wholesalers and merchants know they face a labor problem, and they increasingly see AI as a way to keep productivity rising with a workforce that may be smaller and less experienced.

What many haven’t done is connect those two problems directly.

Where Distributors Get Stuck

Three hard truths, and I’ve watched all three play out.

They buy tools before they capture knowledge. The instinct is to start with a chatbot or forecasting model. But your veteran employee’s know-how is part of the knowledge base AI needs, and if you don’t capture it while that person is still on the payroll, no model can magically recover it later.

In our December 2025 State of AI in Distribution survey, 52% of distributors named people as the biggest barrier to AI: a skills gap at 33% plus change resistance at 19%. Leadership buy-in ranked last. Read that carefully. The executives are increasingly convinced. The organization is struggling to execute.

They frame AI as replacement, and the workforce hears it. Almost no distributor in our data actually expects AI to become primarily a headcount-cutting exercise. Sixty percent expect it to increase the productivity of the people they retain. But if your branch team believes the model exists to replace them, they have little incentive to feed it what it needs.

That creates a dangerous contradiction. The veteran employees whose knowledge you most need to capture may be the least inclined to share it if the project is presented as a way to eliminate jobs.

They wait for clean data and a big platform. Nearly two-thirds of distributors, 63%, are still exploring or piloting rather than scaling AI. The leaders didn’t wait for perfect conditions.

As Grainger Chief Technology Officer Jonny LeRoy put it: “We’ve learned you’ve got to break down your problem into smaller chunks.”

That’s the difference. This is the kind of problem worth working through with people who have already done it, which is a large part of why we built the AI Forum for Distributors in the first place.

What This Means for Your Operation

Reframe the whole thing.

AI is part of your knowledge-retention strategy and your productivity strategy for a workforce that is going to change whether you prepare for it or not. For a UK builders’ merchant, an electrical wholesaler, a European industrial distributor or a multinational distribution group, the terminology may differ, but the operational challenge is the same.

The companies getting this right treat every approaching retirement as a body of knowledge at risk, and they act while the employee is still there to teach the organization.

They point AI first at high-volume, knowledge-dependent interactions: quoting, substitutions, order entry and technical lookups. Email order automation is already the most widely adopted customer-facing AI application in our survey, at 62%, precisely because it handles high volume and its return is relatively easy to see.

The appetite to invest is there. Sixty-five percent of distributors plan to increase AI spending over the next 24 months, with 88% naming productivity as their No. 1 reason for adopting it.

The leaders prove the point. Of more than 300 distributors we analyzed for The AI Execution Gap, only six reached the top AI maturity tier. One of them is Sonepar, the Paris-based global electrical distributor, which has committed more than €2.5 billion to logistics and €1 billion to its Spark digital platform.

The distance between those six and everyone else comes down to execution discipline, not simply budget or software.

What Changes Monday Morning

You can start this week. Five moves.

1. Build a retirement heat map. List everyone within five years of leaving and identify what only they know. That becomes your knowledge-risk map and helps establish your AI priorities.
2. Pick one knowledge-heavy, high-volume workflow. Product substitutions or quoting are strong first targets. Sit with your best person and capture how they do it, including the decisions, exceptions and judgment calls that never made it into the process manual.
3. Start with retrieval, not transformation. Getting technical product information, previous orders and account history in front of a new employee in seconds is a modest, provable win. It also takes pressure off your veterans immediately.
4. Put one senior owner on it. Don’t leave it to IT alone. Our data shows that technology-led efforts can stall when they aren’t tied closely enough to business outcomes. Name a senior person accountable for the result.
5. Measure a commercial number, not activity. Quote turnaround time, first-contact resolution or return rate. If you can’t tie the work to a number a branch manager, managing director or commercial director cares about, it won’t hold.

Notice what’s not on that list: a moonshot, a platform overhaul or a two-year roadmap.

Individually, these moves are modest. Running together, they build the muscle to capture institutional knowledge faster than your people retire.

That’s the game.

Come Work It Through With Your Peers

None of this is theoretical, and none of it is easy to build from a report alone. It is much easier in a room full of leaders wrestling with the same demographic math you are.

That’s what the AI Forum for Distributors, UK and EU is built for. It takes place Oct. 15, 2026, at the National Conference Centre in Birmingham, England, bringing together managing directors and senior commercial, operations, technology and digital leaders from distributors, wholesalers and merchants across the UK, Ireland and continental Europe.

The focus is practical: what’s actually working in distribution, lessons directly from distribution leaders, and a vetted group of technology companies already working in the sector.

Whether your company calls itself a distributor, wholesaler, builders’ merchant, electrical wholesaler or merchanting group, the problem is the same. If some of your best people are within five years of walking out the door, the time to build the capability that preserves what they know is now, not after they’ve gone.

Register for the AI Forum for Distributors, UK and EU, and come build it with people facing the same challenge.

The post Your Best People Are Retiring, But Your Best New Hire Is AI appeared first on Distribution Strategy Group.

That is the gap that matters.

The question is no longer whether AI matters. It is whether you can put it to work in the business. The wholesalers and merchants doing that are beginning to build an operating advantage over those still evaluating the technology.

According to 2025 Boston Consulting Group (BCG) research, companies that systematically embed AI into their operating models can achieve up to five times the revenue gains and three times the cost reductions of laggards. That advantage compounds as leaders lower their cost to serve, improve margins, and put more use cases into production.

The problem is increasingly less about technology than the discipline required to deploy it. Wholesalers that scale AI treat it as an operating discipline owned by the business. Those stuck in permanent pilots too often treat it primarily as an information technology project.

The Execution Gap

The broader numbers show the problem. Recent 2026 reporting indicates 88% of AI pilots never reach production or scale. For generative AI specifically, a 2026 Massachusetts Institute of Technology Media Lab report puts the failure-to-return-on-investment rate at 95%.

That is not necessarily an argument for waiting until the tools improve. It is an argument for fixing the process that turns a working pilot into a daily operation.

The early movers are already visible. DSG research on more than 300 North American distributors found only 26 that met the bar for integrated, production-level AI capability. Grainger, Wesco, Sonepar and Rexel are among the companies that have moved beyond asking where AI fits and are building it into their operating models.

For European wholesalers and merchants, Sonepar and Rexel make the point particularly relevant. This is not simply a North American trend.

European adoption data also shows a significant gap in company size. Eurostat reports that about 20% of European Union enterprises used AI in 2025. Among large enterprises, adoption was 55%, compared with 17% among small enterprises. The larger players are moving faster.

DSG’s State of AI in Distribution research shows a similar divide between individual use and operational deployment. Some 63% of distributors surveyed use ChatGPT, but only 15% have put a proven AI use case into daily operations.

The first number represents people using an AI tool. The second represents companies changing how they operate.

That is the execution gap.

The money is already showing up for companies that moved first. Rexel reported digital sales at 34% of group sales in 2025 and 35% for the first six months of 2026. In Europe, digital represented 44% of Rexel’s sales in 2025, supported by the adoption of digital tools and algorithmic quote and order entry.

That is a European wholesaler building an operating advantage that becomes harder for competitors to close the longer they wait.

Put the Business in Charge

One of the biggest barriers to closing the execution gap is treating AI primarily as an information technology project.

The reflex with new software is often to hand it to the information technology department. But AI changes how work happens at the branch counter, how salespeople negotiate, how buyers procure and how customer service teams process orders.

An information technology leader knows how to connect an application programming interface (API). That person may not have the operating context to redesign the quoting workflow for counter staff.

Without a clear process owner, an accountable executive sponsor and cross-functional involvement, pilots struggle to move into production. The wholesalers and merchants that scale AI put business leaders in charge of business outcomes.

Grainger’s differential investment model assigns high-stakes projects to leaders pulled from operations, people who understand the profit drivers and own the outcome. Sysco elevated its AI strategy to board-level oversight in 2026 and tied it to a $100 million cost-savings target.

The lesson is straightforward: Put accountability with leaders who own the process, the customer, and the financial result.

Start Where the Value Is Obvious

Getting started does not require a companywide transformation program. It means aiming AI at one high-volume, rules-based bottleneck and getting a measurable result inside a quarter.

Several workflows stand out.

Demand forecasting and the bullwhip effect. Wholesale distribution sits in the middle of the supply chain, making it particularly exposed to the bullwhip effect, where relatively small changes in customer demand can create much larger stock swings upstream.

Traditional forecasting models lean heavily on historical sales and moving averages. A retailer batches an order to capture a freight discount, or a manufacturer runs a promotion, and older models can interpret those artificial spikes as changes in underlying demand.

Machine learning models can incorporate point-of-sale information, weather, and economic indicators to identify anomalies and produce a cleaner purchasing baseline. Wholesalers using algorithmic demand sensing report 40% to 50% lower forecast errors, potentially reducing excess stock while improving fill rates.

Start here if working capital is tied up in the wrong stock.

Order entry and margin recovery. Order entry remains one of the back-office functions that can quietly limit revenue capacity. A business-to-business (B2B) wholesaler processing thousands of orders each month through email, PDF files and spreadsheets can still depend heavily on manual keying into an enterprise resource planning (ERP) system.

AI-based order entry can extract line items, map customer part numbers to an item master, validate contract pricing and draft the sales order. Exceptions can be routed to an employee rather than requiring every line to be processed manually.

DSG research puts productivity improvement from automating quote generation at 40% to 70%. Rexel Canada, for example, deployed agentic document processing with Onit and Hyland and achieved near-perfect invoice indexing accuracy within 48 hours in 2026.

Start here if customer service teams are spending too much time on manual order entry.

Pricing and getting salespeople to hold the line. In a high-volume, low-margin business, small improvements in price realization can have an outsized effect on operating profit.

Pricing ranks as the No. 1 AI priority for 27% of distributors surveyed by DSG in 2026. AI pricing engines can analyze customer price sensitivity, purchase frequency, order volume, and competitive position to generate recommendations at the quote level.

But generating the right price is only half the job. The salesperson must use it.

A 2025 study involving a B2B aluminum retailer found that machine learning price recommendations increased profit on treated quotes by 11%. If salespeople routinely override recommendations, however, the technology cannot deliver the intended result.

Start here if margin is leaking through discretionary discounting.

Accounts receivable and cash velocity. Invoice-to-cash has traditionally depended on labor-intensive matching of remittances to open invoices. Accounts receivable automation can issue invoices, ingest payment information, match payments and route disputes based on defined rules.

The models can learn customer remittance patterns and match payments even when reference numbers are missing or a customer short pays.

Start here if days sales outstanding is a constraint.

The Demand Side Is Moving Too

AI readiness is becoming a two-sided issue. Wholesalers and merchants need AI internally to operate more efficiently, but they also need their product and transaction data to be accessible to the AI systems their customers increasingly will use.

Agentic commerce means software acting for a buyer to search catalogs, verify pricing, confirm availability, and potentially execute transactions. B2B procurement, with its repeat purchases, technical specifications and negotiated contracts, is a natural environment for that technology.

DSG’s AI 2030 framework found 61% of organizations expect to deploy fully autonomous AI agents for complex functions within five years.

That changes the importance of product data.

An AI purchasing agent does not need to browse a homepage. It can query structured product information directly. Wholesalers therefore need clean, machine-readable catalogs containing accurate technical attributes, Global Trade Item Numbers (GTINs), and appropriate product classifications.

That also increases the importance of secure APIs that can expose appropriate contract pricing and real-time stock information to authorized systems.

The work required to support machine buyers is another reason to begin addressing product data now rather than waiting for agentic commerce to mature.

Get the Foundation Right Without Waiting

Data is one of the most common reasons companies delay AI deployments, but waiting for perfect data can become its own barrier.

No model will automatically reconcile a duplicated customer master or inconsistent item records. Some wholesalers freeze deployments while waiting for pristine data. Others run algorithms against fragmented ERP information and get confident but incorrect answers.

The better approach is to improve the data required for a specific use case while that use case is being developed.

The European Technical Information Model (ETIM), used across European and North American electrical and heating, ventilation and air conditioning markets, provides a shared product classification structure. In UK building materials, the Builders Merchants Federation (BMF) Product Data Template serves a similar purpose and feeds into Data Yard, the industry data pool developed by the BMF with the National Merchant Buying Society.

Master data cleanup is not something to skip. But it does not have to be completed across the entire business before the first AI deployment begins.

Build Governance in From the Start

European wholesalers also have a regulatory consideration their North American counterparts do not face to the same degree.

The European Union AI Act, in force since August 2024, establishes a risk-based approach to AI governance and can apply to companies outside Europe when their systems affect people in the European Union.

For wholesalers and merchants, some of the clearest requirements involve employment applications such as automated curriculum vitae screening, task allocation, performance evaluation, and workforce monitoring. These can fall into high-risk categories requiring governance, documentation, and human oversight.

AI systems affecting pricing or credit also require careful governance because of potential discrimination risks.

That is not a reason to delay deployment. It is a reason to build governance into the first use case rather than bolt it on later.

What Changes Monday Morning

Moving from evaluation to an operational deployment requires three decisions from the executive team.

Put a business leader in charge. Take AI strategy out of the exclusive control of information technology. Give a senior business leader with profit-and-loss responsibility ownership of the outcomes. Review what workflows changed, what financial return was generated and what process comes next.

Pick one hard bottleneck and commit to a 90-day result. Stop broad, undefined experiments. Pick a specific constraint, such as margin erosion or manual entry of emailed purchase orders, and apply a specialized tool to it. Establish one or two key performance indicators (KPIs) at the start so the business can determine within a quarter whether the deployment worked.

Start the data work in parallel. Do not wait for perfect data. Clean and standardize the information required for the first use case while building the capability to support the next one.

The objective is not to build an AI strategy on paper. It is to put one use case into daily operations, measure the result and use what the organization learns to tackle the next process.

The Two-Quarter Window

The window to treat AI solely as a future capability has closed. The opportunity now is to move it into daily operations.

The wholesalers and merchants building an advantage are not necessarily doing so because they have access to better AI. They are getting better at selecting specific business problems, putting operating leaders in charge, and turning successful pilots into repeatable processes.

The next two quarters provide enough time to do something tangible: Choose a bottleneck, establish ownership, put a use case into production, and measure the result.

Theory does not move market share. Execution does.

AI Forum UK & Europe, Oct. 15, 2026, at the National Conference Centre in Birmingham, will bring together wholesale distribution and merchant leaders to examine AI deployments, implementation costs, timelines, data preparation, and the lessons emerging from putting AI into day-to-day operations.

Register for AI Forum UK & Europe

Where European wholesale distribution leaders turn AI into competitive advantage.

The post The Two-Quarter Window: Why European Distribution Leaders Have to Move on AI Now appeared first on Distribution Strategy Group.

As an Employee Stock Ownership Plan (ESOP), the stakes were personal. Every employee owned shares. When we found a way to handle more without adding cost, the share price got better, and everyone in the company felt it. This was never a slogan about cutting people. It was about not automatically bolting on a salesperson or a customer service representative (CSR) every time the business grew.

Here is the hard part. Most of what we did to grow for free was brutally manual. We rebuilt processes by hand. We fought for every point of margin. It was slow, and it was expensive to figure out. Distributors now have AI tools that make growing for free easier than it has ever been. The reason it works has almost nothing to do with the labor line, and that is the part most business cases miss.

Why Labor Savings Dominate the ROI Conversation

Most distributor AI business cases lead with headcount. Hours saved. FTEs avoided. Cost per transaction. There is a good reason for that. Labor is easy to measure. You know what a CSR costs, you know how long a task takes, and you can multiply your way to a savings figure your CFO will accept.

The easy number is the small number. McKinsey made this point directly in an August 2026 analysis. At most companies, even mature ones, Selling, General, and Administrative (SG&A) runs 5% to 12%  of revenue. A large cut to that line still cannot explain the returns leading AI adopters are reporting. The labor story is real. It is also nowhere near the whole story. Stop your business case at hours saved and you are measuring the least valuable thing AI does.

The Hidden Cost of Slow and Poor Decisions

Here is what almost nobody puts on a balance sheet. Decisions cost money. Not just the labor to make them, but the outcome when they are made slowly or made wrong.

Think about your own operation. A price held a week too long while a competitor moved. A reorder point that lagged real demand, so you either sat on dead stock or ran out. A quote that sat in a queue while the customer got antsy and called someone else. A credit approval that took three days on a deal that needed an answer in three hours. An exception nobody caught until it became a service failure and a phone call from an angry account.

None of those show up as a line item. The daily grind swallows them, invisible, and they are some of the largest operating costs you carry. McKinsey named decision-making one of the largest and least visible costs in a business, embedded in daily activity rather than captured anywhere you can see it. That is exactly why grow for free was so hard for us. The cost we were fighting sat buried in a thousand small decisions, and we had to dig each one out by hand.

The Distribution Decisions Where the Money Actually Sits

Not every decision matters equally. The money sits in a handful of high-frequency, high-consequence calls that your business makes thousands of times a month. In distribution, the list is short and familiar.

Pricing. The right price on the right order, set in the moment rather than off a stale matrix.

Replenishment. Reordering the right SKU in the right quantity before demand shifts under you.

Inventory positioning. Moving stock to where it will sell before a stockout, not after.

Quote prioritization. Working the quotes most likely to close and worth the most, instead of the ones that happen to be on top of the pile.

Credit. Approving good customers fast enough to keep the deal warm.

Exception handling. Catching the order, the shipment, or the account that is about to go sideways while there is still time to fix it.

Every one of those is a decision, made constantly, where speed and accuracy convert straight into margin. That is where the money lives. McKinsey advises companies to start where decisions are both expensive to make and economically important, and points to demand sensing as a natural first move for a distributor. Any operator already knows this in their gut.

How AI Changes Speed, Consistency, and Decision Quality

AI moves three levers at once on these decisions, and they compound.

Speed. McKinsey estimates AI can pull decision timelines from weeks or months down to seconds and cut the cost of deciding by more than 90%. A pricing or replenishment call that used to wait for a weekly cycle can now happen many times a day.

Consistency. Your best decisions stop depending on which person happens to be at the desk. The judgment gets applied the same way every time, at scale, without the good day or bad day variance that comes with any human queue.

Quality. AI can weigh more signals than a person can hold in their head, catching patterns in demand, inventory, and account behavior that a busy team would miss.

Here is the part leaders skip, and it decides whether any of this pays off. McKinsey found that companies layering copilots and dashboards onto their existing processes get only modest financial gains. Companies that redesign the workflow around AI see a 20%  Earnings Before Interest, Taxes, Depreciation, and Amortization (EBITDA) uplift and payback in one to two years. Same technology. The difference is whether you changed how the work flows or bolted a tool onto the old way. We learned that the hard way at Hisco long before AI showed up. You cannot grow for free by speeding up a broken process. You rebuild it first.

How to Measure a Decision Dividend

If the value sits in decisions, hours saved is the wrong scoreboard. McKinsey borrows a better one from the factory floor. You judge a plant by what it produces, not by how many machines are running. Apply the same test to AI. The measure is not how many tools you licensed. It is how many of your decisions are now informed, accelerated, or automated by AI. McKinsey calls that decision throughput and argues it is the metric that matters most.

Track it where it counts. What share of your pricing decisions are AI-informed today versus a year ago? How fast does a quote move from request to response now? How quickly does a credit approval clear? How many stockouts did you catch before they happened rather than after? Those numbers tie directly to margin and to revenue, and they tell you far more than a tally of hours saved ever will.

Where Human Approval Still Belongs

Faster and more decisions are not automatically better decisions. A high-speed engine bolted to a bicycle just wrecks faster. The discipline is knowing which decisions AI can make on its own and which still need a human in the loop.

McKinsey is blunt about this, and so am I. You define, clearly, when AI can act on its own, when it escalates to a person, and who owns the result. Skip those rules and two things go wrong. Either leaders refuse to delegate anything and throughput dies, or the system runs unchecked and the first bad call in a high-stakes lane turns into a real problem. Routine reorders and standard pricing are where AI should run. The large, unusual, or relationship-sensitive calls are where your experienced people keep the final say. Drawing that line is a leadership decision, not a technical one, and it belongs to the business, not to IT.

How Decision Automation Changes Management Roles

When AI takes over the routine decisions, the manager’s job changes. It shifts from making every call to setting the rules the system follows, watching the exceptions, and improving the logic over time. Less doing, more governing. That is a genuinely different job, and the people who thrive in it are not always the ones who were fastest at the old one.

This is where the veteran gets more valuable, not less. AI holds the information. Your twenty-year people hold the meaning, the context, the read on why this account or this pattern breaks from what the model sees. The edge comes from putting those together. The company that wins is not the one that bought AI. It is the one that built an organization able to turn better decisions into better results, week after week.

We chased the decision dividend at Hisco before anyone had a name for it, the slow and manual and painful way, and watched it show up in a share price every employee owned. The goal was never to shrink the team. It was to grow without the cost growing alongside it. Distributors now have tools we would have paid anything for back then, and most are pointing them at the wrong target, shaving minutes off tasks instead of sharpening the decisions those tasks feed.

Here is where to start Monday. Pick the one decision your business makes most often that moves margin the most, pricing on a common order type is a good candidate, and time it end to end. That number is your baseline and your first AI use case. Build from there.

What is the most expensive decision your business makes every single day, and how long does it take you to make it?

The post The Decision Dividend: Why AI ROI in Distribution Is Bigger Than Labor Savings appeared first on Distribution Strategy Group.

I keep thinking about that model as I watch distributors approach AI. Because most of them are doing the opposite. 

What Sysco Actually Did 

Sysco put its AI strategy under formal board oversight this month. The board renamed its Technology Committee the Artificial Intelligence Transformation and Technology Committee, and it now meets monthly with management to oversee adoption. The company tied that governance to a $100 million cost-savings program in its fiscal 2027 outlook, driven by AI-based process improvements and automation. 

The reaction I expect from a lot of small and mid-size distributors: that’s Sysco. $84 billion in sales, 333 distribution centers, 75,000 employees. Of course they can stand up a board committee. We can barely get our team to agree on which chatbot to use. 

I understand the instinct. I’ve run operations where every initiative fought for the same three people’s time. The instinct is still wrong, and it’s the kind of wrong that costs you two years you won’t get back. 

The lesson from Sysco isn’t the dollar figure. It’s the decision to treat AI as an operating strategy with an owner and a cadence, instead of a pile of disconnected projects nobody is accountable for. That decision has nothing to do with your revenue. A $60 million distributor can make it on a Tuesday. Most won’t because they’re waiting for permission, they think only scale provides. 

Where Most Distributors Get Stuck 

When they finally decide to put someone in charge of AI, they hand it to IT. It feels natural. AI is technology, IT owns technology, done. 

That’s backwards. 

I’ve spent the last two and a half years working to understand how AI can impact distribution, and I’ve looked at all of it through the lens of someone who spent years in the trenches. Here’s what that time convinced me of. The people who understand the business are the ones who must guide the strategy. The IT people are the ones who help execute it. Get that order wrong and the whole effort tilts toward what’s technically interesting instead of what moves the operation. 

I love IT. The good ones are remarkable at making systems talk to each other and keeping the whole operation running. That’s exactly the problem. They’re busy keeping the business operational. AI adoption is not a systems integration job first. It’s a set of decisions about how work gets done at the field level, where customer interactions happen. Which processes change. What you stop doing. Where the savings land. 

The tools themselves are amazing. That’s not in question. But a tool without context fails to hit the objective. Your IT leader can tell you whether an AI tool connects to your ERP. Your IT leader cannot tell you whether rewriting your quote-to-order flow is worth the disruption to your counter team, because that call requires knowing what happens when a customer calls in a rush order and the rep must make a judgment. That’s business knowledge. It lives in the people who’ve worked the field, not the server room. 

What Two and a Half Years Taught Me 

When I started digging into AI for distribution, I assumed the hard part would be the technology. It isn’t. The technology works. The hard part is knowing where to point it, and that judgment doesn’t come from a demo. 

Watch what happens when a distributor buys a capable AI tool and hands the rollout to someone without operational depth. The tool gets configured to do what the vendor’s example showed, not what the business needs. It answers questions nobody was asking. It automates a step that wasn’t the bottleneck. Six months later the honest verdict is that it technically works and nobody uses it. 

Now watch the same tool in the hands of someone who’s run a branch. They know the rush order is where margin leaks. They know the counter rep’s judgment call is the moment that keeps or loses the account. They point the tool at that moment, shape it around how the work really flows, and the thing starts paying for itself. Same software. Completely different result. The difference is entirely the context of the person guiding it. 

That’s the pattern I’ve seen again these two and a half years. The tool is rarely the variable that decides success. The person deciding how it fits the business is. 

Why the Grainger Model Matters More Now 

This is where the Grainger differential investment model matters more than ever. Put an upcoming leader from the business on AI. Someone who understands how the company operates, who’s watched customer interactions go right and wrong, who can look at a tool and see where it fits in the actual flow of the work. That person guides the strategy. IT helps them execute it. Run it in that order and you’ll succeed where a technology-led project stalls, because the context the tool can’t supply on its own is sitting in the driver’s seat. 

There’s a moment from Grainger that makes this concrete. When the company wanted to understand how our sister operation Acklands-Grainger ran itself as a business, we didn’t send executives to study it from thirty thousand feet. We didn’t send IT. We assigned a group of people who understood how the operations worked and how sales happened. That’s what made the difference. They could see the real mechanics, not the org chart version, because they’d lived the same work themselves. 

That’s the same call you’re making with AI. The person who understands the operation sees where a tool fits and where it breaks. Send the wrong person to figure it out and you get a report that reads well and changes nothing. 

Sysco understood this. They didn’t route AI to the CIO and walk away. They put it in front of directors and tied it to operating targets. They made it a business job with business accountability. 

The Objections I Hear, and What I Tell People 

Two pushbacks come up every time I make this argument. Both deserve a straight answer. 

The first comes from IT leaders, and some of it is fair. Plenty of IT leaders are business-fluent. They’ve sat with sales, they understand the customer, they’d run this well. If that describes your IT leader, then you already have the person. My point isn’t that the title says IT. It’s that the person needs deep business context, wherever they sit on the org chart. What you can’t do is default the job to IT because AI has the word technology attached to it. Default assignment is the failure. Deliberate assignment to the right person is the win. 

The second comes from smaller distributors. You may not have a spare general manager to reassign. I get it. You’re running lean and everyone already carries a full load. But this isn’t a full-time job at your size. It’s one person, a few hours a week, and a standing monthly meeting with you. If you can’t free up that much for the single technology shift most likely to reshape your cost structure this decade, that’s worth sitting with. The distributors who find the hours now are the ones who won’t be scrambling to catch up in two years. 

What the Owner Actually Does 

Naming an owner isn’t the finish line. It’s the start. Here’s what the role looks like in practice, so this doesn’t become another title with no teeth. 

In the first 90 days, the owner does three things. They map where AI could move a number in your business, not where it looks impressive, but where it touches revenue, cost, or a customer experience that’s costing you. They run one or two focused pilots against those spots, small enough to move fast, real enough to matter. And they set up how you’ll know it worked, in dollars or hours or retained accounts, before the pilot starts, not after. 

From there, the monthly meeting carries the weight. Every month the owner reports three things to you: what changed, what it saved or earned, and what’s next. If a pilot isn’t working, it gets killed or fixed, not quietly carried. If one is working, it gets resources to scale. IT sits in that conversation as the execution partner, building the integrations and keeping it stable, working against a direction the business has already set. 

That’s the whole structure. An owner with business context, a concise list of high-value targets, honest measurement, and a monthly meeting where decisions get made. It isn’t complicated. It’s just disciplined, and discipline is the part most distributors skip. 

What Changes Monday Morning 

Here’s what I’d tell any distributor who isn’t Sysco. Name one person from the business who owns AI. Not a committee. Not IT by default. A general manager, an operations VP, a rising leader who understands your customers and carries a number. Give them a standing monthly meeting with you where they report what changed, what it saved, and what’s next. Let IT execute against that direction. That’s the whole governance structure. It costs you nothing but the discipline to hold the meeting. 

The distributors who install that ownership now will spend the next two years compounding real operational gains. The ones who park AI in the IT backlog will spend those same two years running disconnected experiments that never add up to anything. 

Sysco showed you the play. The barrier to running it isn’t your size. It’s whether you’re willing to name an owner from the business and hold the meeting. 

Do not miss any content from Distribution Strategy Group. Join our list.

The post Sysco Just Made AI a Board-Level Job. You Don’t Get to Say You’re Too Small appeared first on Distribution Strategy Group.

Customers who ordered from us every week suddenly went quiet. No complaint, no lost bid, no phone call. The orders just stopped. When we dug in, we saw something evolving more. A competitor had put someone on site, or stood up onsite replenishment, and the account was gone before we knew it was in play. The reason wasn’t a coincidence. As organizations got leaner, they were looking for ways to keep their people in the building instead of running to a branch, waiting at a counter for material, and hauling it back. Onsite replenishment gave them that. By the time the drop-off showed up in our numbers, there was nothing left to save.

What lost us those accounts wasn’t price and it wasn’t service. It was a change in how the customer wanted to buy, and we found out about it too late to respond. The buying process moved, and we were still standing where it used to be.

What’s coming with AI agents is that same silent loss, with one difference that makes it worse. With onsite replenishment, I could eventually walk into that customer’s facility and see my competitor’s rep restocking the shelves. The cause was visible once I went looking. With an agent, you won’t watch it bypass your site for another supplier. You won’t get a shot at winning the deal back. The agent will read your product data, find it wrong or incomplete, and move on to a supplier whose information is clean. You’ll never know the agent was there. The agent saw everything.

For 20 years, ecommerce in distribution meant one thing. A human went to a website, searched for a part, compared a few options, and clicked buy. Every version of that process, from paper catalogs to punch-out to modern web stores, was built around a person doing the browsing. The whole apparatus, the product photography, the merchandising, the search bar, the account login, assumes a set of human eyes on the other side of the screen.

That assumption is breaking.

AI agents are starting to do the shopping. Not recommend. Not assist. Shop. An agent gets a task from its owner, a procurement manager, a facilities director, a contractor, and it goes out and finds the part, compares specs and price across suppliers, picks a winner, and starts the purchase. No human looks at your website. No human reads your product page. No human ever sees your brand. The buyer set the intent. The machine does everything after that.

This isn’t a forecast for 2035. The pieces are already in the field. Buyers are running research through chatbots. Procurement teams are testing agents that pull quotes across suppliers. The technology to let an agent complete a purchase end to end exists now. What’s still forming is how fast your specific customers adopt it, and that timeline is not yours to control.

What the Navu Data Actually Showed

At our AI forum in Atlanta, John Greely from Navu put numbers to a shift most distributors feel but haven’t measured. More than 80% of searches that return an AI overview now end without a click to any website. Ninety percent of B2B buyers research before they ever talk to you. Half of B2B software buyers start their buying journey inside an AI chatbot, not on Google.

Read those three numbers together and the picture is clear. Fewer visits. Higher intent on the ones you get. And a growing share of buying decisions that get made before a human ever reaches your site, or without a human reaching it at all.

Greely made a second point that matters just as much. The distributors still measuring their website by raw traffic are measuring the wrong thing. Traffic volume was the right metric when the goal was getting found by as many people as possible. That goal is fading. When a smaller number of higher-intent buyers, and their agents, come to your site, every visit carries more weight than the one before it. The question stops being how many people showed up. It becomes whether the ones who did, human or machine, got what they needed.

The website’s job used to be getting found. Now the job is answering. And the thing doing the asking is increasingly not a person.

Why This Is Different from Every Ecommerce Shift Before It

Distributors have lived through disruption before. Catalogs went to websites. Phone orders went to online orders. EDI and punch-out wired us into customer procurement systems. Marketplaces showed up and took a slice of the transaction. Each of those changes moved the buyer somewhere new, but the buyer stayed human. You could still influence the decision with a relationship, a rep who knew the account, a service reputation, a clean website. The human on the other end could be persuaded, reminded, recovered.

An agent removes the human from the search entirely. It doesn’t care about your relationship. It doesn’t remember that your counter team saved the customer’s job last spring. It reads structured data, checks price and availability and spec compliance, and it decides. If your product information is incomplete, out of date, or hard for a machine to retrieve, you don’t get a lower ranking. You get skipped. The agent never surfaces you as an option, and the buyer never learns you were one.

That’s the part worth sitting with. The old failure was losing a comparison. The new failure is never being in the comparison. When a human shopped your site and left, you at least had a chance at analytics, a retargeting ad, a follow-up call. When an agent evaluates you and rejects you, there’s no bounce to measure and no cart to recover. The rejection happens inside a system you can’t see, based on data you may not know is wrong.

It’s the same trap as those onsite replenishment losses, moved into software and sped up. Back then, at least a rep could eventually notice the account had gone cold and go find out why. An agent doesn’t leave that trail. The account just quietly stops showing up in the pipeline, and the explanation lives in a data quality problem nobody flagged.

A Scenario Worth Picturing

Picture a regional electrical distributor with a solid book of business. Good branches, loyal counter customers, a website that works fine for the humans who use it. One of their larger industrial accounts brings in a procurement agent to manage routine reorders, the repeat-buy items the customer purchases every month without much thought.

The agent’s job is simple. For each item on the reorder list, find a supplier with the right spec, in stock, at a competitive price, and place the order. It checks this distributor’s site along with three others. On most items the distributor is competitive. But on a handful of SKUs, the website shows a lead time that’s three weeks stale, and on two others the spec fields are blank because that data never got filled in after a catalog update. The agent can’t confirm those items meet the requirement, and it won’t guess. It routes those lines to a competitor whose data is complete.

Nobody at the distributor sees this happen. There’s no lost bid in the CRM, no angry phone call, no RFQ that came in and went out. The monthly reorder volume from that account just slowly pace is lower than prior and. If anyone notices, the likely first assumption is pricing, and the sales team goes chasing a discount that was never the problem. The problem was three stale lead times and two empty spec fields that a machine read, judged, and acted on in under a second.

That scenario is illustrative, not a named account. But every piece of it is happening in the field right now in some form, and the failure mode is exactly the kind of thing that hides in plain sight until the numbers force the question.

What You Actually Control

None of this means the distributor is helpless. It means the point of leverage moved. It used to sit in the sales conversation and the customer relationship. A lot of it now sits in your data.

Three things decide whether an agent can find you and choose you. First, your product information must be complete and correct. Specs, dimensions, compatibility, availability, price. The stuff a human counter rep fills in from memory is exactly the stuff an agent needs on the page because the agent has no counter rep to ask. Second, that information must be structured so a machine can retrieve it cleanly, not buried in a PDF, locked behind a lead-capture form, or trapped in an image nobody tagged. Third, it must be current. An agent working from your stale availability data will quote a customer a lead time you can’t meet or skip you for a competitor who shows real stock. Wrong data doesn’t just cost you that order. It teaches the agent your site can’t be trusted, and agents don’t forget.

That’s not a marketing project. It’s an operational one. And it belongs to the same people who already own product data, inventory accuracy, and pricing discipline. The work that makes you visible to an agent is the same work that makes your counter faster, your quotes more accurate, and your customers less likely to get a surprise on a lead time. You’re not building a new capability from scratch. You’re finishing the data work most distributors have been putting off for a decade.

What Changes Monday Morning

Start by asking a question you can answer this week. If an agent searched your category today, would it find you, and would the data it found be right? Pull up your ten highest-volume SKUs and look at them the way a machine would. Is the spec information complete? Is availability accurate? Is price current? If the answer is no on your best-selling items, it’s worse everywhere else, and everywhere else is where the long tail of your margin lives.

Then look at where your product data lives. If the real answers sit in a rep’s head, a supplier PDF, or a spreadsheet nobody syncs, an agent can’t use any of it. Map the gap between what your best people know and what your website can prove. That gap is your exposure. Every item in it is a line an agent might route to a competitor because it couldn’t confirm you were a fit.

Third, change what you measure. If your website scorecard is still built on sessions and pageviews, you’re grading yourself on a test that’s being retired. Start tracking whether high-intent visits, human or machine, find complete and accurate answers. Greely’s framing is worth stealing outright. Ask what the top ten questions were that your buyers, and their agents, asked last month, and whether your site answered them.

The buyers who still show up as humans will forgive a thin product page. They’ll call the branch, and your counter team will save the day like it always has. The agent won’t call. It will read what’s there, judge it, and move on. Silently. The distributors who get ahead of this aren’t the ones with the flashiest site. They’re the ones whose information is complete, structured, and current, so the machine can trust it, and choose it.

This is the conversation we’re having at the Applied AI for Distribution Conference. If you’re trying to get ahead of the agent-driven buyer, that’s where to be. appliedaifordistributors.com

Join DSG in Birmingham: Distribution Strategy Group will bring its Applied AI for Distributors Forum to the U.K. and Europe on Oct. 15, 2026, at the National Conference Centre in Birmingham, England. The one-day event will bring together distribution executives and AI leaders to examine how distributors are putting artificial intelligence to work across sales, operations, customer service and other parts of the business. Learn more and register at Distribution Strategy Group’s AI Forum UK & EU

The post When the AI Agent Becomes the Customer: What Distributors Risk Losing appeared first on Distribution Strategy Group.

As I progressed in my career and moved to other distributors, something clicked. The products we sold at Grainger? They were the same products we were selling at my new company. Same manufacturers. Same specifications. Same availability.

What was different? The experience we brought to customers. And that’s become the only real differentiator in today’s world.

The Evolution from Transactions to Touchpoints

Moving up through operations and into leadership roles, my perspective shifted again. It became less about individual transactions and more about how we were delivering experience across all the touchpoints in our business. That’s the real win in any customer experience work you do.

You need to understand all the points where your customer touches your business and know how each one is performing. Here’s the thing that keeps executives up at night: you can have an amazing experience up front with placing an order and receiving a delivery, but if you have a not-so-great experience on the back end when the customer wants to pay the invoice, all the work you’ve done up front has completely fallen apart because that one touchpoint failed.

I’ve seen it happen. A distributor invests in improving quote turnaround times, trains their counter staff to be more responsive, and optimizes their delivery routes for speed. All excellent work. Then customers hit accounts receivable with a billing question and suddenly they’re waiting three days for a call back or dealing with an inflexible credit policy. One weak link in the chain undermines everything else.

The Two Ways Customer Experience Initiatives Fail

I’ve lived through both extremes of how companies approach customer experience measurement, and both miss the mark.

At Grainger, we did surveys every quarter. The consistency was there. That’s because we measured religiously. But much of what we got back wasn’t actionable. We’d see scores, we’d track trends, but we didn’t have clear direction on exactly what to fix or where to focus. Valuable information, collected regularly, but not translating into concrete operational improvements.

At another company I worked for, we went the opposite direction. We’d do surveys occasionally when someone decided it was time. No consistent rhythm. No follow-through. And we shouldn’t have been shocked that we never got any better. We weren’t measuring consistently, and when we did measure, the insights weren’t specific enough to drive action.

Both approaches fail for the same fundamental reason: they’re missing the continuous improvement loop. It’s not enough to measure frequently if you don’t know what to do with the data. And it’s not enough to get actionable insights if you only measure once and never verify whether your changes worked.

The Measurement Cycle That Actually Works

Real progress comes from a complete cycle: measure what matters to customers at each touchpoint, analyze the results to identify specific operational gaps, implement targeted improvements based on those findings, then measure again to verify those changes moved the needle. This sustains the gains you’ve made. It’s continuous. It’s actionable. And it’s the only way to systematically improve customer experience rather than just tracking it.

Just today, I was reviewing results with a distributor who’s been following this approach. 18 months ago, their Net Promoter Score was 61—already superior performance by industry standards. But they didn’t rest on that. They continually and systematically measured to get better. They identified specific gaps between what customers valued and how they were performing, made targeted operational adjustments, measured again to confirm improvement, and then built those improvements into their standard operating procedures. Their latest score? 73. They went from good to exceptional by treating experience as a continuous improvement process, not a one-time achievement.

That kind of sustained improvement isn’t unusual when distributors commit to the full cycle. We typically see 12% to 18% improvement in customer satisfaction scores within the first year for companies that measure consistently, get actionable insights into what specifically to fix, implement those changes and measure again to track progress.

But here’s where many distribution companies stumble: they make improvements and see scores rise, then assume the work is done. Six months later, performance slides back. Why? Because they didn’t sustain the improvements. The cycle isn’t measure-analyze-improve-done. It’s measure-analyze-improve-sustain-measure again. You need to lock in the gains by updating training materials, revising standard procedures, and continuing to monitor performance so improvements become permanent rather than temporary fixes.

The pattern is consistent: identify the specific touchpoints where performance lags what customers care about, make concrete operational changes, verify those changes moved the needle, build them into your ongoing operations, then start the cycle again. Not measurement for measurement’s sake. Not occasional surveys that gather dust. A real loop that drives lasting improvement.

One distributor told us recently that their customer satisfaction metrics have become “our barometer of what to do.” Not a nice-to-have data point filed away somewhere. The actual guide for resource allocation and operational priorities—measured consistently, acted on specifically, sustained through process changes, and verified through the next measurement cycle.

The Real Product Sitting on Your Shelf

Walk into any distribution warehouse and you’ll see rows of products. But talk to the customers who keep coming back, and they’ll tell you something different. They’re not buying your ball bearings. They’re buying the fact that when their production line goes down at 4:30 on a Friday, you answer the phone. They’re buying the reality that your inside sales team knows their operation well enough to catch a potentially wrong order before it ships. They’re buying the seamless experience from quote to delivery to invoice.

They’re buying every interaction they have with you.

This isn’t just intuition. When you measure what drives customer loyalty, asking them to rate not just their overall satisfaction but the importance and performance of specific touchpoints like delivery precision, quote turnaround, credit flexibility, and billing accuracy—patterns emerge. The companies that excel at the handful of things customers genuinely care about across the entire journey. They keep those customers. The ones that excel at one or two touchpoints but fail at others? Well, price becomes the tiebreaker.

What This Means For Your Operation

Here’s where most distribution companies get stuck. They invest millions of dollars in inventory, hundreds of thousands of dollars in warehouse automation, significant capital in fleet vehicles. All critical investments. But then they treat customer experience like an afterthought—something the customer service department handles when there’s a problem.

That’s backwards.

Your inventory management system tells you exactly how many units of each stock-keeping unit (SKU) you’re carrying. But can you tell me with the same precision how long customers wait on hold? How many times must the average buyer call to get an order update? How satisfied they are with your billing process? Whether your credit terms align with what matters to them?

The distributors winning in competitive markets treat customer experience with the same rigor they apply to inventory turns and fill rates. They measure it systematically tracking not just whether customers are satisfied overall, but which specific capabilities matter most to them at each touchpoint and where performance gaps exist. They get actionable insights that point to concrete fixes. They implement those improvements. They sustain those changes by embedding them into standard procedures. Then they measure again to verify progress holds.

A distributor might discover their Arizona branch has slow quote turnaround times that don’t exist in California. They address it by revising the quoting workflow. Six months later, they measure again to confirm Arizona’s performance improved. Then they update training materials and performance metrics to sustain the improvement. Or they find that construction customers rate them lower than manufacturing customers on delivery precision. They adjust delivery processes for construction accounts, train drivers on the new standards, and track whether satisfaction moved—and stayed there.

These aren’t massive strategic problems requiring complete overhauls. They’re specific, fixable issues at individual touchpoints that directly impact whether customers stay or leave—and the only way to know if your fixes worked and stuck is to keep the measurement cycle going.

The Interchangeable Product Problem

When products become commodities, purchasing behavior shifts. Price matters, but it stops being the only thing that matters.

A purchasing manager facing identical products at similar prices will choose the distributor that makes their job easier across the entire transaction. The one that provides accurate order tracking. The one whose team responds to emails within an hour instead of a day. The one that handles returns without an interrogation. The one whose billing is straightforward and whose credit team understands their business cycles.

Your competition isn’t just other distributors anymore. It’s Amazon Business setting expectations for same-day delivery transparency. It’s consumer experiences training buyers to expect real-time updates and frictionless transactions at every step.

The gap between what customers experience in their personal lives and what they tolerate in business-to-business (B2B) transactions is closing fast.

What Changes Monday Morning

Stop treating customer experience as something you understand intuitively and start measuring it with the same discipline you apply to financial metrics. But don’t just measure—make sure you’re getting actionable insights that tell you specifically what to fix. Ask customers what matters most to them at each stage of doing business with you, then track how you’re performing on those specific dimensions across every touchpoint.

Then—and this is the part most companies skip—do something about what you learn. Make targeted operational adjustments based on clear priorities. Build those changes into your standard procedures so they stick. And measure again in six months to see if those changes moved the needle and held. That’s the continuous improvement loop that works.

The insights won’t require a complete business transformation. More often, they’ll point to specific operational adjustments at touchpoints that have outsized impact on retention. It’s integrating customer feedback directly into your customer relationship management (CRM) system, so your sales team sees it in real time. It’s identifying that your Milwaukee customers are genuinely satisfied across the board while your Phoenix customers love your sales team but struggle with inventory availability. It’s discovering that your accounts receivable process is the weak link undermining otherwise robust performance.

These aren’t abstract improvements. They’re concrete changes that protect revenue—but only if you have consistency in measurement, actionable insights that tell you what specifically to improve, and the discipline to sustain those improvements through process changes and ongoing monitoring.

Your products are increasingly interchangeable. Your experience across every touchpoint doesn’t have to be. But you need the complete cycle: measure consistently, get actionable insights, implement specific improvements, sustain those gains and measure again to verify progress holds.

 Are you measuring customer experience consistently enough to track real trends? Are your measurements telling you specifically what to fix? And when you make improvements, are you building them into your operations so they last?

That’s the difference between a measurement program and a continuous improvement system that protects revenue.

The post Your Product Isn’t What Customers Are Really Buying appeared first on Distribution Strategy Group.

I was lucky enough to begin my career at Grainger, and they taught me early that the lens of the customer is the most important one you have. What I learned there, and with every company since, is that looking at the business through the lens of the value stream is how you make improvements that matter.

At one of those companies, I watched a good account walk out the door, and not one department thought it was their fault. Sales had hit quota on that account. Customer service had closed every ticket inside its service window. The warehouse posted a 98% fill-rate that month. Credit has done its job protecting us from a slow-pay risk. Every scoreboard reads green. The customer still left and took 11 years of purchases with them.

The bottom line: customer experience is a company-wide responsibility that crosses every function you run. It must be a team effort because no other way works. Most distributors manage it as if a single team can own it. That’s backwards. And it’s why so many improvement efforts stall after the first survey.

The Myth of Customer Experience (CX) Ownership

The instinct, when a leadership team decides customer experience matters, is to name an owner. Hand it to marketing because they run surveys. Hand it to customer service because they answer the phones. Create a customer experience (CX) manager and check the box.

Back at Grainger, we relied heavily on cross-functional teams, and the person in charge usually wasn’t from customer service. It was the person who could impact that problem the most. Their job ran past their own department. They had to bring the rest of the departments along, so that when we implemented a solution it didn’t break something else in the value chain. That distinction matters.

Here’s the reality of our industry. There’s rarely a formal customer experience (CX) title. Marketing usually spearheads the data because they own the survey. You can even name an owner. None of that changes what the work is. Whoever you put in charge inherits responsibility for an outcome and authority over almost none of the inputs. They can’t set credit policy. They can’t change the warehouse slotting that drives short ships. They can’t rewrite how sales set delivery expectations. So, they do the one thing within reach: they send more surveys, build prettier dashboards, and watch the number sit flat, while the actual drivers of dissatisfaction live in the other departments that never got the memo.

You can assign a steward for customer experience. You cannot delegate it. Those are different things and confusing them is where most distributors get stuck.

Why Customer Experience (CX) is a Team Sport

Walk one order through your building and watch how many hands touch the customer.

Marketing sets expectations before the customer ever calls. Sales make the promise on price, availability, and delivery. Customer service fields the question when something’s unclear. Operations and the warehouse pick, pack, and ship it, and decide whether it arrives complete and on time. Inventory and supply chain decide whether the item was even there to sell. Finance, through credit and billing, decides whether the order ships today or sits on hold, and whether the invoice is clean or triggers a dispute. Executive leadership decides whether any of these groups get measured on the customer’s experience or only on their own departmental number.

Lay your touchpoints out on a wall and you’ll count a lot more boxes than people. They don’t map one to one. Eight or nine names end up owning all of them, and which names matter depends on the touchpoint you decide to fix. The divisions are also finer than the organizational chart admits. Inside finance alone, the credit manager who puts an order on hold isn’t the billing clerk who lets a bad invoice go out. Same department, two different players, two different ways to lose a customer.

The customer doesn’t see eight or nine departments. They see one company, and they judge you on the weakest link in the chain. A flawless sales relationship doesn’t survive, a billing department that fights every credit. A great price doesn’t survive, a backorder nobody communicated.

The Problem with Departmental Thinking

Now it gets dangerous. Every department optimizes the metric it’s measured on, and each one looks like a winner while the customer’s experience erodes.

Say you’ve got a credit problem. The worst thing you can do is hand it to the credit manager and tell them to fix it. They will. They’ll fix it for credit, optimize it for their number, and in the process create a problem for sales, for service, for inventory. The fix is local and the damage is company wide. Multiply that across every function: credit tightens terms to protect days sales outstanding (DSO), and a customer who’s bought from you reliably for a decade gets treated like a flight risk. The warehouse hits its fill-rate target by shipping the easy lines complete and shorting the one item the customer built their job around. Sales books the order and moves on, never flagging that the delivery date was optimistic. Each manager defends their number on Monday. Each number is real. The customer is still unhappy, and no single report shows why.

This is the part operators understand from the plant floor: you can run every workstation at peak efficiency and still ship a bad product, because the problem lives in the handoffs, not the stations. Customer issues always originate in one department and surface in another. The credit hold becomes the service team’s angry phone call. The slotting decision becomes the salesperson’s lost renewal. Look at only department by department and you’ll never find the root cause, because the root cause is the seam between two departments that don’t share a scoreboard.

A Value-Stream Approach to Customer Experience

Stop managing customer experience as a set of departments and start managing it as a value stream. A value stream is the full end-to-end sequence of activities that carries a customer from first request to delivered product—the whole path, not any one department’s piece of it. It’s the same discipline you’d apply to any operational process improvement, pointed at the customer instead of the warehouse.

One of the companies I worked with made every one of us own a value stream. Not in your department. A value stream that cut across all of them. Owning it meant owning the whole path—the handoff coming into each department, the work that happened inside it, and the handoff back out to the next one. You had to walk into credit, into the warehouse, into billing, into customer service, and learn what each group did, how they did it, and how their piece landed on the customer. It was uncomfortable, and it was the most useful thing I did that year, because it forced me to see the handoffs instead of the boxes. That’s what real ownership of customer experience looks like: somebody who can shepherd a problem all the way through the organization rather than optimize one stop on it. Doing it this way forces you to think of the customer first and see it through that lens.

Map the full journey the way the customer travels it: discovery, quote, order, credit approval, fulfillment, delivery, invoicing, support, reorder. Then ask three questions at every step. Who owns this touchpoint? What does the customer expect here? And how do we know whether we’re delivering it? Most distribution leaders can’t answer the third question with data at more than half the steps. That gap is the whole problem.

A value stream needs shared visibility. Everyone must see the same customer, the same feedback, the same metric, at the same time. The credit manager needs to see that the account they just put on hold is one your top rep has been nurturing for two years. The warehouse needs to see that the line they shorted last week is the reason a customer rated delivery a two. When the data sits in silos, nobody owns the seams, and the seams are where you lose customers.

How the Cross-Functional Team Operates

A cross-functional team lives or dies on how it’s run. I won’t tell you what your org chart should look like, because a single-branch distributor and a national platform don’t share one. The mechanics, though, travel everywhere. Five rules separate a committee that meets from a team that moves.

Pick the lead by leverage, not title. The person who can move the problem the most runs the team, whether that’s the credit manager, the ops lead, or a service supervisor. When the problem changes, the lead changes.

Charter it around the value stream, not a department. The mandate is the customer’s path through the issue, start to finish. The lead doesn’t fix their own piece and hand it off. They carry the fix across every department it touches.

Put everyone on one scoreboard. While the team is working, every member is measured on the customer outcome, not their own departmental number. Drop that rule and the credit manager goes right back to protecting days sales outstanding (DSO).

Give it decision rights and a clock. The team meets on a fixed cadence, makes calls that cross department lines, and escalates the minute it hits a wall it can’t clear. A team that needs permission for every cross-functional move die of slowness.

Disband it when the seam is fixed. Then stand up the next one around the next problem. The capability is permanent. Any single team is temporary.

Get those five right and you stop coordinating departments. You start moving as one company toward the customer.

How Technology Enables Cross-Functional customer experience (CX) Management

This is where good intentions die for a practical reason: you can’t shepherd a problem across eight or nine departments if you can’t see the customer in one place. Spreadsheets, an annual survey, and a gut feel won’t get you there. You need a dedicated customer experience platform built to do the cross-functional work, not just collect feedback.

A platform earns its place when it does three things your spreadsheets can’t. It centralizes feedback so every department reads from the same source instead of trading anecdotes. It surfaces the ownership gaps, the touchpoints where the customer is struggling and no one’s accountable. And it shows you how much each capability matters to the customer, not only how you score on it, so you invest where it moves the relationship instead of where it’s easy.

That’s the design philosophy behind Customer Experience RX, the platform we built at DSG specifically for distributors. It puts importance and performance side by side on every capability, benchmarks you against other distributors, and lets you slice feedback by segment, geography, and job function in one portal every department can open. One distributor used it to target the improvements that mattered most to their customers and moved their Net Promoter Score from 57 to 70 in a single year.

What matters is what the tool makes possible: one version of the truth is that finance, sales, service, and operations all trust enough to act on together. A platform won’t fix your customer experience. It gives your leadership team the shared visibility and accountability to fix it themselves.

The Leadership Takeaway

Customer experience is the main differentiator left in distribution. Product lines converge, prices get matched, and the company that wins is the one the customer trusts to get it right across the whole relationship. That’s a leadership problem before it’s a software problem.

What changes Monday morning:

Stop asking which department owns customer experience. Own it yourself, at the leadership level, and build the cross-functional team to carry it, led by whoever can impact the problem most and bring the other departments along. Map the value stream. Put one source of customer truth in front of all eight or nine stakeholders. Measure importance alongside performance so you invest in what customers value. Then do the unglamorous work of fixing the seams between departments, because that’s where your customers are quietly deciding whether to stay.

Your competitors are still arguing about whose fault the last lost account was. Get your teams looking at the same customer, and you’ll stop having that argument. That’s the game.

The post Customer Experience Is Everybody’s Job appeared first on Distribution Strategy Group.

Fill rates, inside sales productivity, pricing decisions, and customer service. That’s what you were managing last week. That’s the job.

But something happened recently that deserves your attention because it has direct implications for how your operation runs.

The U.S. government ordered a major AI model provider to restrict access to one of its models over national security concerns. The directive arrived late on a Friday. Because cloud providers cannot easily separate foreign nationals from domestic users, the company did the only thing it could do: It shut the model off for everyone. Live sessions ended in errors. New requests were quietly routed to an older, less capable system. Users were never asked. The provider made the decision and executed it automatically.

Here’s the part distributors should pay attention to: You don’t use that model directly. Almost nobody does.

But you use Copilot. Your customer relationship management platform added an AI assistant. Your quoting platform drafts proposals. Your software vendor launched a chatbot.

Many of those features rely on a large language model underneath, the same category of technology that was just shut off. If that model becomes unavailable, you’re not the one who chose to depend on it. Your vendor did. You may not realize there’s a problem until something stops working. And because the rest of the software may continue functioning normally, the root cause can be difficult to identify.

That’s the real issue: The dependency is hidden one layer down.

The AI tools spreading fast through distribution are not standalone AI subscriptions. They’re business applications with AI features embedded inside them. An enterprise resource planning module that drafts purchase orders. A customer service bot that answers questions at 2 a.m. A natural language search box on a dashboard.

In most cases, the AI is a feature, not the product itself. As a result, few distributors stop to ask what is powering it.

It’s also important to be precise about where the risk exists.

Not every AI feature in your software carries this exposure. Your demand forecasting system may rely on statistical models that have been in place for years. Your pricing engine may use machine learning that predates the current generative AI wave entirely. Those systems are unaffected by the type of disruption described above.

The exposure is concentrated in generative AI features, the tools that write, summarize, chat, and answer questions in plain language. Those capabilities typically depend on one of a small number of large language model providers.

And for those features, your vendor made the decision.

They may have built their application around a single provider. They may switch providers without informing customers. You don’t know unless you ask.

Distributors manage this type of risk every day in their physical supply chains. You know which products are sole sourced. You know which suppliers lack a viable backup. You would never allow a critical stock-keeping unit (SKU) to depend on a single supplier without understanding the risks.

Yet that’s exactly what many distributors are doing with AI-enabled software today because nobody asked the right questions when the software was implemented.

Here are the questions worth asking now.

Which features use a large language model, and whose model is it?

Separate generative AI features from everything else. The assistant and chatbot rely on a large language model. Forecasting and pricing algorithms do not. For every feature that depends on a large language model, determine whether the vendor relies on a single provider or multiple providers.

If the model is open source, which model is it? What is the country of origin of the model?

Open source is not a single thing. Open weights (Meta’s Llama, Mistral) give a vendor real optionality; open documentation without infrastructure access gives them a marketing bullet point. Also, country of origin matters: models from China-domiciled organizations carry different regulatory exposure than European or American alternatives.

If the model is their own trained model, is your data used to train it?

Ask this specifically: does my data improve your model, and does that improvement benefit other customers? Vendor contracts routinely permit training use under “service improvement” language broad enough to include your pricing logic, purchasing patterns, and margin structure. If the answer is yes, you are funding their product development with your operational intelligence—potentially including the model performance your competitors will use against you next year. Consider a data processing addendum that explicitly excludes training use, or price that contribution accordingly.

What happens if that model becomes unavailable?

Can the vendor switch to another model, or does the feature stop working entirely? Vendors that have thought through this scenario will have an answer. Vendors that have not are effectively transferring their architectural risk to you.

How quickly can you recover, and how will you know there’s a problem?

If the underlying model goes dark, are you facing an outage measured in hours, days, or weeks? Will the vendor proactively notify customers, or will your sales team discover the quoting assistant has stopped working in the middle of a shift?

What’s your exposure if you lose the tool for 60 days?

This question is for your team, not your vendor. For every AI-enabled application performing meaningful work, understand what breaks and identify the manual fallback process.

The good news is that most AI-enabled software is not yet mission critical. That gives distributors a window to ask these questions while the answers are still relatively inexpensive.

That window will not stay open forever.

As AI becomes more deeply embedded in distributor operations, these tools will move closer to the same level of importance as enterprise resource planning systems, procurement platforms, and key supplier relationships. Most distributors won’t think about these dependencies until a disruption occurs.

The ones that ask the questions now will be better prepared when someone else’s model suddenly goes dark.

Many AI-enabled features inside enterprise resource planning, customer relationship management and sales platforms depend on third-party large language models that distributors neither selected nor control. Understanding those dependencies now can help prevent unexpected operational disruptions later.

The post You Don’t Know Which Artificial Intelligence Models Are Running Inside Your Software. That’s the Problem appeared first on Distribution Strategy Group.

Every operator learns this lesson eventually. There are people you give tasks to, and there are people you hand projects to.

Task people need the work broken down for them. Pull this report. Call these three customers. Chase that PO. They execute, but you carry the thinking. The work moves only as fast as your attention.

Project people are different. You hand them an outcome and a deadline. They build the plan, work the steps, check their own output, and come back when it’s done or when they hit a wall that genuinely needs you. Every operation I ever ran got better the day I found one more project person. They’re rare, and they change what a leadership team can take on.

Until now, every AI tool you’ve touched has been a task person. A fast one. A tireless one. But a task person. You ask, it answers, you carry the thinking.

That just changed. Anthropic, the company behind the Claude AI assistant, released something new this week called Fable 5. It’s the first widely available AI built to work like a project person. You don’t need to follow the technology news to care about this one. Here’s what it is in plain terms, and why it matters to your business.

What Actually Got Released

Set aside the product names and version numbers. Here’s the simple version.

Anthropic built the most capable AI it has ever made. It was powerful enough that the company didn’t feel comfortable handing the unrestricted version to the public. So, they released it two ways. The full-strength version goes only to a small group of vetted security organizations and key infrastructure companies such as money center banks. The second one is Fable 5; a public version is the same brain but with a safety harness on it. You and your team can use that second version today.

What makes it different from the AI you’ve already tried? Two things.

First, it can work on something for hours or days, not seconds. The AI tools most of us know give an answer and stop. This one takes an assignment, builds a plan, works through it step by step, checks its own work, and keeps going until the job is done.

Second, it can read your documents the way a person does. Not just the words. The charts, the tables, the numbers buried in a price file or a contract. It can look at a page and understand what’s on it.

One caution: it costs more to run than the everyday versions. This is not the tool for quick questions and email drafts. The cheaper tools handle that fine. This is the specialist you bring in for the hard jobs.

The Safety Story Your Board Will Ask About

Why did Anthropic hold back the full-strength version? Because in early testing, it was good enough at things like finding security weaknesses that the company got cautious. That should tell you something about how capable this generation is.

The public version handles this with a built-in checkpoint. When someone asks it to do something in a sensitive area, the system catches the request and hands it to a less powerful AI instead. You don’t manage any of this. It’s automatic.

Why should a distribution executive care? Because it changes the governance conversation — though not in the way you might assume. The built-in checkpoint is a real feature, but it is not proof that Fable 5 is safer for your company’s data. It may be it may also carry risks no one can fully assess yet. Nobody has enough experience with this generation to say for certain.

So don’t treat the safety harness as a reason to relax. Treat it as a reason to look closely. Before this tool goes anywhere near sensitive data, put a designated team member — or a small group — in charge of evaluating how it handles your information and where the exposure points are. Let the people who own your data security form a view first, then decide how far to take it.

Three Things It Can Do That Are Worth Money

First, you can hand it whole projects. The work that soaks up your senior people’s bandwidth is rarely task work. It’s the pricing review. The branch network analysis. The product line cleanup that’s been “next quarter” for six quarters. This AI can take an assignment like “go through three years of sales and inventory data, recommend a stocking strategy by branch, and show me the tradeoffs” and work it end to end.

Second, it can read what your business runs on. Distribution lives in vendor price sheets, contracts, rebate schedules, spec sheets, and system exports. Most of that has been invisible to AI until now. This one can read those files directly, pull exact numbers out of tables and charts, and connect what it finds across documents. Years of information sitting in your shared drives just became usable.

Third, it can build and connect software. Most distributors carry a backlog of system projects: getting the enterprise resource planning (ERP) system to talk to the sales system, building a customer portal, creating internal tools for pricing and quoting. Those projects stall because technical talent is scarce and expensive. This AI is unusually good at exactly that work, which means the backlog starts moving.

Notice what’s missing from that list: chatbots. The value here isn’t conversation. It’s completed work.

Where to Start

Don’t roll this out to everyone. That’s the mistake distributors made with the last wave of tools, and it produced a lot of activity and very little outcome.

Instead, pick one or two projects that have been sitting on a senior leader’s desk for months because nobody has the bandwidth. The analysis you know you need but keep deferring. Assign one to this tool to one or two people who have proven themselves to be adept, competent users of AI tools and run it as a 90-day pilot. Measure two things: how fast you got to a decision-ready answer, and how that answer stacks up against what your team would have produced.

Keep the everyday AI tools doing the everyday work. Use an expensive tool for expensive problems. The math only works when the project touches the P&L: pricing, working capital, network design, system projects.

And put your best operators on the review side. The AI has information. Your veterans have context. The output is only as good as the judgment evaluating it.

Every previous wave of AI made individuals faster. This one changes what leadership can delegate.

The Real Shift

That moves the constraint — and this is the shift to sit with. The limiting factor is no longer what the technology can do. It’s whether you know which projects you’d hand it. Most distributors can’t answer that question, because they’ve never written down the backlog of high-value analysis they’re not doing. The work that would move margin or working capital but never gets staffed.

That backlog is now addressable. Your competitors’ backlog is too.

In the meantime, here’s the Monday morning question: if you could hand an entire project to a tireless analyst tomorrow, do you know which one you’d pick? If the answer takes you more than a minute, that’s the work to do first.

Brooks Hamilton, A.I. Strategy Advisors, also contributed to this article.

The post The New AI Just Changed What You Can Hand Off appeared first on Distribution Strategy Group.

The room gets quiet. There’s usually a pilot somewhere. A chatbot on the website. Sometimes a pricing tool that’s technically AI-powered but nobody’s touched the configuration in six months.

That’s not an AI strategy. That’s a press release.

93% of wholesale distributors tell us AI is a priority. 16% have moved past exploration. I’ve been staring at that gap for 18 months, and it’s not closing fast enough. The companies inside that 16% aren’t just ahead on a capability. They’re building a compounding advantage that it gets harder to close every quarter. Grainger, WESCO, Sysco, and a handful of others are deploying AI across pricing, inventory, sales, and warehouse operations at the same time. The rest of the industry is still drafting the roadmap.

That gap is the story. And DSG is going to name it.

Introducing the AI Top 50

Starting June 23, we’re publishing the most rigorous ranking of AI maturity in wholesale distribution that’s ever been produced. The AI Top 50 evaluates 55 companies across 34 data points per organization: deployed use cases by function, technology stack depth, executive commitment signals, measurable outcomes, and evidence quality. No surveys. No self-reporting. Every data point sourced and verified from public evidence, executive statements, company announcements, and direct research.

The result is a four-tier ranking that tells you, with specificity, where every major wholesale distributor sits on the AI maturity curve. And exactly what separates a Tier 1 company from a Tier 4.

The methodology is defensible. Ian Heller and I built it to hold up under scrutiny from a CFO, a technology analyst, or a skeptical board member. We scored what companies have deployed, not what they’ve announced. We measured outcomes, not intentions. We cited every claim.

Two Phases. Six Months. One Franchise.

We’re releasing the AI Top 50 in two phases.

Phase 1, the AI Top 25, drops June 23 at the Applied AI for Distribution Conference in Chicago. The Top 25 covers Tier 1 and Tier 2 companies: the distributors who’ve moved from exploration to execution, deployed multiple AI applications, and can point to measurable results. This is the prestige release. Conference attendees get the full 30 to 40-page report, an interactive dashboard filterable by tier, vertical, and company size, and a Google NotebookLM knowledge base they can query in natural language during the conference itself.

Phase 2, the full AI Top 50, expands to Tier 3 and Tier 4 companies in October, timed to the UK AI Forum. That’s where the picture gets complete. Who’s leading. Who’s gaining ground? Who’s still on the sideline with a stated commitment and no deployment evidence.

The two-phase structure is deliberate. It extends the content lifecycle from a single event into a six-month franchise. It creates a second news cycle in the fall. And it gives the companies sitting on the bubble between positions 20 and 30 a specific reason to have a conversation with us about what would move them up.

A Ranking Creates Accountability. That’s the Point.

The distribution industry has been remarkably tolerant of the gap between what leaders say about AI and what they’ve done. Part of that is cultural. Distribution is a relationship business and public comparisons feel uncomfortable. Part of it is that until now, there hasn’t been a rigorous, public benchmark that named names.

The AI Top 50 changes that. It gives distribution executives a clear external reference point for their board conversations, their technology investment decisions, and their competitive positioning. A CEO who reads that her company is Tier 3 while two of her primary competitors are Tier 1 doesn’t need a consultant to tell her what to do next. The data makes the case.

That accountability is a feature, not a side effect.

The Research Behind the Ranking

The 55 companies we evaluated represent the full spectrum of AI maturity in wholesale distribution. The Tier 1 cohort, the 13 companies at the top, includes large-cap distributors with dedicated AI teams, measurable ROI across multiple functions, and technology stacks built around platforms like Proton.ai, Blue Yonder, GAINS, and Zilliant. These are the companies setting the pace.

Tier 2 covers 14 active adopters who have deployed two or more AI applications and are expanding. Tier 3, 22 companies, are in early-stage adoption, running pilots or single-function deployments. Tier 4, six companies, have expressed clear AI intent with limited deployment evidence to back it up.

One thing I want to be clear about. Being in Tier 3 or Tier 4 right now isn’t a permanent condition. It’s a snapshot of where a company stands in May 2026. The distributors who move between Phase 1 and Phase 2, who close the gap between where they are and where the leaders are, will have done the real work. That’s who we want to track.

June 23. Chicago. Be in the Room.

The AI Top 25 goes live June 23 at the Applied AI for Distribution Conference. If you’re a distribution executive trying to understand where your organization stands against the industry’s AI leaders, that’s where you want to be. Not to hear about it secondhand. Not to read the summary three weeks later. To be in the room when the rankings drop and ask the questions that matter to your operation.

Registration is open at appliedaifordistributors.com. The report will all be available live for conference attendees.

Ninety-three percent committed. Sixteen percent executing.

Which number does your company belong to?

The post DSG to Reveal First AI Top 50 Ranking of Wholesale Distributors appeared first on Distribution Strategy Group.