Most CEOs Buy AI Licenses. WD-40 Bought Operating Margin.

THE CRAFT

Somewhere in your house right now, there is a blue-and-yellow can of WD-40 in a kitchen drawer next to the duct tape and the spare batteries. You have not thought about WD-40 in months. You will not think about WD-40 again until something stops working — a sticky garage door, a squealing hinge, a stuck bolt — at which point you will reach for the can and expect it to be there. If it is, the moment is invisible. If it is not, you buy something else next time.

That is the only customer interaction WD-40 has. There are no follow-up emails, no loyalty programs, no community-management AIs, no quarterly business reviews. There is one transaction: was the can on the shelf the moment the customer reached for it?

For thirty years inside WD-40 Company, that one transaction has been failing in ways nobody outside the building noticed. Not catastrophically — the brand is too iconic for that. But quietly, in distributor markets where the local planning system was running on a homegrown spreadsheet stack, where seasonal forecasts were built by humans against macro shocks they couldn’t see coming, where a Brazilian inflation spike or a Middle East container disruption would translate, three weeks later, into a hardware retailer somewhere in the Americas being out of stock on the SKU the customer was actually reaching for. And underneath that — the planners, the people whose job was to prevent those gaps, were spending their Mondays running manual reforecasts against last week’s broken assumptions instead of doing the work they were actually hired to do.

Steve Brass watched both of those problems for thirty-one years.

He joined WD-40 in 1991, twenty-something, a UK national with a French and German degree, and took a job at the European commercial office in Milton Keynes. European Commercial Director. Division President for the Americas. Chief Brand Officer. President and Chief Operating Officer. In September 2022, when Garry Ridge retired after twenty-five years, Brass took the CEO seat.

The first thing he did was not hire McKinsey, build an AI strategy deck, or call a single technology vendor. The first thing he did was take a public position on two outcomes WD-40 had been failing for thirty years. The customer outcome: the can will be on the shelf the moment the customer reaches for it. The employee outcome: my planners will spend their days solving exceptions and shaping strategy, not running manual reforecasts.

Only after both positions were taken did the technology procurement begin. The order matters more than any single piece of software that followed.

This is the AI story I think a small-cap or mid-cap operator needs to read on a Sunday night. Not the boardroom strategy deck story. Not the four-use-cases-and-an-AWS-bill story. The story of a CEO who knew, before he ever booked a vendor demo, exactly which customer outcome and which employee role were going to be the targets of the work. And what happens when the answer to “which AI tool should we buy?” is the third question instead of the first.

THE OPERATOR

The situation

WD-40 Company is the rare public company where you can describe the entire revenue base in a paragraph. The flagship product — WD-40 Multi-Use, the blue-and-yellow can — accounts for the majority of sales. There is a smaller higher-margin specialty line called WD-40 Specialist, plus a few household-cleaning brands the company picked up over the years. Three operating segments: Americas, Europe-Middle East-Africa, and Asia Pacific. Most of the actual selling happens through distributors — hardware stores, automotive aftermarket retailers, big-box, industrial supply.

The operational reality, until Brass took the seat, was three decades of distributor-market improvisation. Each market ran on whatever planning system the local team set up when they came in — often a homegrown spreadsheet stack, a regional ERP, paper-based reorder logic. The planning department, at the global level, was a group of human planners running a manual workflow. Sales history. Seasonal trends. Distributor purchase-order patterns. Best-guess forecasts that worked fine until something broke.

Two specific failures, year after year, in the same shape:

The customer failure. When a macro shock hit — a tariff, a container-cost spike, a Middle East supply disruption — the manual forecast couldn’t adjust fast enough. Three weeks later, a hardware retailer somewhere in the distributor network was short on the WD-40 Multi-Use SKU. The customer who reached for the can found an empty shelf, bought a competitor, and the brand absorbed the loss as “the cost of doing business.” The financial cost was measurable in stockouts and dead inventory. The brand cost — every customer who bought a substitute and then realized it worked fine — was not on any P&L line.

The employee failure. Every Monday morning, the planners were running manual reforecasts against the previous week’s broken assumptions. They were the people inside WD-40 who knew most about distributor behavior, channel dynamics, and seasonal pattern shifts. They were spending most of their week typing those insights into spreadsheets that would be obsolete by Wednesday. The work that would have made the most difference — exception management, distributor relationship work, scenario planning against actual macro shocks — was the work they had no time to do.

Both failures had been visible to anyone with thirty years inside the building. They were two halves of the same problem. The customer was being failed because the planner was being misemployed. The planner was being misemployed because the system was treating manual reforecasting as the primary task instead of as the lowest-impact part of the job.

That is the scene in San Diego on the morning Steve Brass took the CEO seat in September 2022.

The move

Brass executed in a specific sequence. Each move was an operational decision first, a technology procurement second.

First, he took the customer position publicly, on the record. On the Q2 fiscal-2026 earnings call (April 9, 2026), Brass framed the strategy in language that does not appear in most CPG earnings calls. The technology was discussed only after the customer outcome was named. The line from the call that locked the company into a direction: “Our goal isn’t just personal efficiency; it’s rethinking processes across the business.” That is a sentence about replacing the way work happens — for customers and for employees — not about giving people copilots. The market has been told, on a public earnings call, that processes are being rebuilt. You cannot walk that back. The technology procurement that came afterward was already locked into a job by the position Brass had just taken.

Second, he named the specific customer-failure moment as the deployment target. Out of the hundreds of operational gaps inside WD-40, Brass picked one: the distributor stockout under macro shock. The forecasting failure that produced empty shelves three weeks downstream of an external disruption. Then he funded the technology that would close that specific gap — a planning solution called JustEnough, originally built by Mi9 Retail and later acquired by ToolsGroup, sitting underneath a ToolsGroup machine-learning layer that analyzes historical sales, seasonal trends, and external market signals. The job description was clear before the vendor was selected: detect the macro signal, regenerate the forecast in hours instead of weeks, push the corrected order to the distributor before the shelf goes empty. ToolsGroup was the execution. The customer position was the strategy.

Third, he took the employee position — that the planners would be reassigned to higher-impact work, not augmented with copilots and not laid off. This decision came before any ERP technology was selected. Brass’s planners would no longer spend their Mondays on manual reforecasting because the ML layer was now doing that work. Instead, they would do exception management, distributor relationship work, and scenario planning. That redefinition of the job was the second strategic decision. Then Brass extended the ERP backbone — Microsoft Dynamics 365 — across the U.S., Latin America, Asia distributor markets, and parts of Canada, sequencing rollout by where the manual handoffs were costing his team the most. About half of WD-40’s global revenue now sits on Dynamics 365. In Q2 fiscal-2026, the company went live with another phase of the Canada rollout. The remaining half of the revenue base — primarily EMEA — is still on legacy. That is not a missed deployment. It is a sequencing logic that goes employee-burden first, technology-coverage second.

Fourth, he layered the supporting systems around the new operating model. Salesforce became the customer-facing system that gives the reassigned planners visibility into distributor relationship dynamics. Atlas, a supply-chain platform built by John Galt Solutions, is queued behind Dynamics 365 and ToolsGroup as the next-stage planning layer for global supply chain. In WD-40’s own framing, Atlas is “yet to be rolled out globally” — it rolls out as the underlying data feeds get clean enough to support it. None of these tools were procured first. Each one slotted into a role the operating-model redesign had already defined.

The thing to notice about this sequence is what is not in it. There is no Chief AI Officer. There is no “AI Center of Excellence.” There is no four-quadrant strategy slide naming “use cases across the value chain.” Brass has not told the market WD-40 is becoming an “AI-first” or “AI-native” company. He has told the market that the customer outcome is being defended and the employee work is being elevated — and that the company will be more profitable as a result. The AI is the means. The two positions are the strategy.

The result

The customer-side outcome is starting to be visible — and the financial receipt is in the margin line. The Q2 fiscal-2026 quarter (reported April 9, 2026) posted gross margin of 55.6%, up 100 basis points year over year from 54.6%. Operating income climbed 13% to $26.3M on sales growth of 11%, which means operating margin expanded too. Adjusted diluted EPS rose 14%. WD-40 Specialist sales — the higher-margin industrial specialty SKUs — grew approximately 20% year over year, which specifically tells you the can was on the shelf in the higher-margin channels at the moment the customer reached for it. Management explicitly attributed the margin lift in part to the systems investments.

That last sentence is the one to read twice. Most mid-cap CPG CEOs are spending five to fifteen million dollars on AI licenses right now. WD-40 spent on the same kinds of tools — JustEnough, ToolsGroup ML, Dynamics 365, Salesforce, Atlas — and got 100 basis points of gross margin expansion in return. The vendors are the same. The order was different. The margin is the receipt.

The real test is still ahead. WD-40’s CFO has told investors the supply-cost impact from Middle East conflict will not fully land until Q4 of fiscal 2026, given the 90-to-120-day lag from raw-material cost into product cost. The planning AI is the system that has to defend the customer outcome when the shock arrives. Q2 was the easy quarter. Q4 is the one Brass built the deployment for.

The employee-side outcome is harder to measure publicly but easier to feel inside the building. The planners have been reassigned. Manual reforecasting has been industrialized — done by ToolsGroup ML in hours, not by humans in weeks. The planner’s job is now exception management and strategic scenario work. That is the change Brass referenced when he said “rethinking processes across the business” on the earnings call. It is, in operational terms, the most consequential employee-experience shift inside WD-40 in a generation. It is also why the financial signal is showing up in the higher-margin Specialist line — because the people who were hired to know distributor relationship dynamics are finally spending their week doing that work.

The financial signal lags both of those operational shifts and is the third-tier proof point, not the headline. Q1 fiscal-2026 missed forecast. Q2 fiscal-2026 beat. The volatility is what an AI deployment looks like when the cost of deployment hits before the operational benefit shows up. The signal that the deployment is working will not be the headline EPS — it will be the customer position holding through Q4 as the supply-cost shock from Middle East geopolitics arrives. That is the metric the work was designed to defend.

What Brass has not done — and this is worth saying out loud — is give a single TED talk, McKinsey keynote, or HBR column about WD-40’s “AI transformation.” There is no manifesto. There is a press cycle in trade publications, an investor presentation, and an earnings call in which the framing was deliberately understated. Brass appears to have made a deliberate choice to underclaim the strategy publicly and let the financial results do the talking. That is unusual for a CPG CEO. It is also, structurally, exactly what an inside-out CEO with thirty-one years of operational context would do — because the audience he cares about most is not the AI press, it is the analysts who price the stock and the distributors who fulfill the orders.

The Craft of AI read

Here is what I think is going on, and why it matters for a small-cap or mid-cap operator reading this on a Sunday night.

The dominant AI-deployment story in mid-market CPG right now is the outside-hire change-agent CEO. A new leader comes in, surveys the operation, hires McKinsey or Accenture, builds the AI strategy deck, names the four use cases, funds the program, and eighteen months later announces a “transformation.” The company has spent $5–15 million and produced three abandoned pilots, one chatbot the call center didn’t ask for, and a press release.

Brass did the opposite of that, and the opposite is the lesson.

Here is the load-bearing point of the whole piece, and the only sentence in this issue I would want a reader to memorize: Brass did not buy AI licenses. He bought a position on what was good for the customer and what was good for the employees. The licenses came after — and the operating margin came after that. If he had procured ToolsGroup and Dynamics 365 without taking those two positions first, he would have spent five million dollars to make the planning department slightly faster at the wrong work, and the margin line would have shown nothing.

That is the difference between an AI deployment that defends the business and an AI deployment that becomes a line item the next CEO has to write off. Tools cannot decide what work matters. Only the operator can do that. And the operator can only do it by taking explicit positions — on the customer outcome being failed and the employee role being misused — before any vendor enters the conversation.

There is a sequence underneath this that applies to any AI deployment, in any business, at any scale. WHAT before HOW before WITH WHAT. WHAT outcome is being changed — for customers and for employees? HOW does the work need to be redesigned to produce that outcome — what gets industrialized, what gets elevated, what gets eliminated? Then, and only then, WITH WHAT technology do we execute the redesign? Most CEOs run that sequence in reverse. They start at WITH WHAT — they pick a tool, write a procurement memo, and try to retrofit a strategy around it. That is the failure mode every McKinsey-driven AI engagement of the last three years has produced.

There is a second-order point that is specific to a brand like WD-40, and worth lifting out for any operator running an iconic single-product business. AI’s most defensible role inside a 70-year-old single-product brand is not growth. The customer doesn’t want an AI experience with WD-40. They want the can to be on the shelf when the moment of need hits. That is the customer outcome. It is also a position. Most legacy CPG CEOs cannot articulate it that cleanly, because they are still trying to convince their boards that AI is going to remake the business. Brass made a different bet: the business does not need to be remade. It needs to be defended at the moment of customer need and elevated at the moment of employee use of judgment. The same AI deployment serves both — because the planner who is no longer doing manual reforecasts is the planner who keeps the can on the shelf.

For an operator running an iconic specialty product — a category-leading household chemical, a regional grocery house brand, a specialty food, a machinist tool — the WD-40 lesson is not that you need an AI strategy. It is that you need to take two explicit positions, in this exact order: which customer outcome are we refusing to keep failing, and which employee role are we refusing to keep wasting. Find those two positions. Write them down. Tell your board, your team, and your distributors what they are. Then the technology procurement will be obvious — because the tool’s job description is already written.

The thing I would have done differently if I were Brass: I would have separated out the planning-AI narrative from the broader Dynamics 365 ERP rollout in earnings disclosure. Right now those are bundled in management commentary as “investments in productivity-driving enhanced systems.” If Brass split out the planning-AI line — the way Lindsay Corporation will eventually have to split out FieldNET subscription ARR — the analysts pricing the stock would have something cleaner to put in their model. That is a cosmetic complaint against a real piece of work. But cosmetics and disclosure shape multiples.

Things to consider

• Take both positions before you take any vendor call. The customer position (the outcome you refuse to keep failing) and the employee position (the work you refuse to keep imposing on your team) are the two strategic decisions that lock the rest of the work into place. Brass took both publicly, on the record, in language that could not be walked back. If you are funding AI quietly, without naming the customer outcome and the employee role you are protecting, you are getting none of the lock-in benefit and you are also handing the framing to whichever competitor takes a public position first. The vendors will sell to either of you. The market will only remember the one who said it out loud.

• Sequence the rollout by employee burden, not by region size. Half of WD-40’s global revenue is on Dynamics 365 and half is still on legacy. That is not a deployment failure — it is sequencing logic. Brass deployed first into the markets where the planning team’s manual handoffs were costing the most operational time, not into the markets with the largest revenue. The principle for an operator with twelve distributor relationships: rank your distributors by how much their order cadence costs your team’s calendar each quarter, then deploy AI in that order. Most operators get this wrong by deploying alphabetically, by region size, or by where the IT department has bandwidth. Order it by employee burden. The financial result follows.

• Replace the function. Do not augment it. Most AI deployments inside CPG companies right now are “copilots” — tools that sit alongside human planners, marketers, customer-service reps, and supposedly make them more productive. Brass is doing something different. The ML layer is replacing the manual-reforecasting function. The human planners are not getting copilots; they are getting reassigned to exception management and strategic work. This is structurally a more durable AI deployment than a copilot, because copilots are easy to abandon and hard to measure; replacements are hard to abandon and easy to measure. If your AI deployment cannot be reduced to “this function is being industrialized so the people can be elevated,” it is probably a copilot, and it is probably going to fail to land.

• Underclaim publicly. Let the operating result do the talking. Brass has not given a single TED talk, McKinsey keynote, or HBR column on WD-40’s “AI transformation.” There is a press cycle in trade publications, an investor presentation, and a deliberately understated earnings-call line. The first instinct of most CEOs deploying AI is to over-narrate — because the narrative is supposed to lift the multiple. Brass is letting the customer outcome (the can on the shelf) and the employee outcome (planners doing higher-impact work) lift the multiple. That is what an inside-out CEO with thirty years of context does — he trusts the audience that prices the stock to recognize the result without being told. For a small-cap or mid-cap operator who does not have a public audience yet, the corollary is even stronger: spend the narrative budget telling your team and your distributors what the customer and employee positions are. They are the audience that makes the strategy real.

• The result lags the work by 18–24 months — and the metric to watch is the operational one, not the financial one. Q1 fiscal-2026 missed forecast. Q2 fiscal-2026 beat. The volatility is what an AI deployment looks like when the cost of deployment hits before the operational benefit shows up. The signal that the deployment is working will not be the headline EPS. It will be the customer position holding through Q4 of fiscal 2026 as the supply-cost shock from Middle East geopolitics arrives — meaning the customer position (the can on the shelf) was defended and the employee position (planners doing exception work, not manual reforecasting) produced the right call. For a small-cap or mid-cap operator funding an AI deployment right now, do not measure it against the headline P&L for the first eighteen months. Measure it against the customer outcome it was built to defend. If that outcome holds under shock, the financial story will follow.

THE WORKBENCH

Do this on Tuesday

Block sixty minutes on your calendar. Take three sheets of paper.

On sheet one, write the customer position. Name the single moment where a customer reaches for your product and you are not there — the equivalent of WD-40’s empty hardware-store shelf. For a specialty food brand, it might be a grocery shelf during a promotional window. For a regional services firm, a Friday afternoon when a competitor’s response time was fifteen minutes faster. For a specialty manufacturer, a customer’s urgent reorder that took your team a day longer to confirm than it took the competitor. Be specific. Who reached, when, where, with what alternatives available. Then write a single sentence: “This is the customer outcome we refuse to keep failing.” That sentence is the position.

On sheet two, write the employee position. Name the work your most knowledgeable people are spending their days on that they should not be spending their days on. The manual reforecasting equivalent inside your business. The spreadsheet that gets rebuilt every Monday. The reports nobody acts on. The handoffs that cost your senior people three hours a week each. Be specific. Who is doing the work, what would they do instead if the work disappeared, what would the business gain from that reassignment. Then write a single sentence: “This is the work we refuse to keep imposing on our team.” That sentence is the second position.

On sheet three, write the AI product that closes both gaps. Not a forecast. Not a dashboard. A directive. A single specific instruction for a single specific person, generated automatically the moment the upstream signal is strong enough to act. “Ship two extra cases to this distributor by Wednesday.” “Open a second pick wave at this DC.” “Push back the promo by 48 hours.” The product description has to make clear what the customer outcome being defended is, what the employee work being eliminated is, and what the human is now free to do instead.

What you will learn: the strategic work is taking the two positions. The technology procurement is downstream and almost easy. Most of what passes for AI strategy at your scale is four half-baked attempts at growth-side AI products, none of which earn their keep against the cost of the customer absences they failed to prevent and the employee time they failed to liberate.

What it costs you if you don’t do this: another year of customer absences and employee misallocation that you absorb as “the cost of doing business” — when half of each were preventable by an operating model you could have decided on this Tuesday.

The rigorous version

The Tuesday exercise gets you two positions and one AI product description. It is the shape of the work, not the work.

The rigorous version is The Ground-Up Workshop. It is how you and a small group of your subject-matter experts get on the same page — for two days in person — about what AI is actually for inside your business. Not a strategy deck. Not a vendor selection. A worked-through customer position, a worked-through employee position, and the operating-model sketch that makes both possible.

Every AI strategy you’ve been sold starts with vendors. This one starts with the small group of people in your business who already know where the operational truth lives. Two days in person. A target operating model that has AI built in from the ground up. A 90-day starting plan. The alignment step that gets everyone on the same page about what AI can actually do for your business — before the deeper work begins.

What you walk out with: a customer position you can defend in writing, an employee position your team can recognize themselves in, a target operating model with AI embedded at the work level (not at the org chart), and a 90-day starting plan that aligns the team on what to do first. The deeper operational work — the build, the procurement, the rollout — comes after, if and when you’re ready. The workshop is how we start.

Price: $10,000. I run a small number of these each quarter. What you walk out with is the document Brass would have wanted in 1995 — the customer position the AI is built to defend, the employee position it is built to liberate, and the operating-model sketch that makes the technology procurement obvious.

THE QUESTION

What is your customer position — the moment you refuse to keep failing? What is your employee position — the work you refuse to keep imposing on your team?

Most CEOs cannot answer either question, because they have not spent thirty years inside the company. They are external hires, and the customer-failure pattern was never theirs to break. The employee misallocation was never theirs to feel. They will hire McKinsey. They will fund four use cases. They will get three abandoned pilots and an AWS bill.

Brass spent thirty-one years watching the same customer moment fail and the same employee work get wasted. The day he took the seat, he knew which two positions to take. Then the technology procurement was obvious. Then the operating margin showed up. The order is the lesson — and the margin line is the receipt.

You either know the two positions, or you don’t.

If you want to talk through what the two positions look like for your company — the customer outcome you have been failing, the employee role you have been wasting, the operating model that fixes both — hit reply, or send a note to grant@thecraftofai.com. I run a small number of these conversations each quarter, and I start each one by asking you the same two questions I just asked above. If you can answer them cleanly, we can probably do real work together. If you can’t, the workshop is designed to get you to the answers.

— Grant grant@thecraftofai.com