2025 ComEx Lessons Learned

Around two dozen commercial, transformation and supply chain leaders taking stock of the year and setting priorities for 2026. No featured speaker and no prepared material — a loose agenda of what worked, moderated by John Follett. Contributors are described by role and industry, under the Chatham House Rule.

The short version

• Adoption and adaptation are separate problems. Most people had adopted AI superficially — using a chatbot as a search engine — while the way they work stayed where it was.
• The AI that earned its budget produced something a competitor lacks. Meeting summaries were described in the room as almost as irritating as writing them.
• A pivot from margin optimization to revenue growth reaches KPIs, segmentation, risk tolerance, deal rules, the customer story and the incentive plan. It is a rebuild rather than a change of emphasis.
• Over-engineering was the year's self-diagnosis: hundreds of bespoke KPIs, layers on layers, and programs trying to change everything at once with the resources to change one thing.
• Training is not enablement. What worked was role modeling, guided discovery, and staying with people while they build confidence.
• Tariffs rewarded the companies that had mapped the worst case, stood up a cross-functional office, and read their contracts before the increase landed.

Digital dexterity, and being honest about your own

The opening lesson came from a device and digital innovation lead in pharma who had taken two medical devices through FDA approval during the year. His takeaway was not about the regulatory path.

Product innovation gets the attention and the plan; the business model innovation beside it gets overlooked. Shipping, handling, pricing and route to market all had to be worked out for something the market had no prior category for, and each step turned out to be a mountain climb rather than a footnote. Behavior change and business model change, he said, are far harder than the product launch itself.

He then turned the same lens on himself and the room, and the phrase he used stuck for the rest of the session.

A global head of commercial excellence in pharma had run the same problem to ground with five whys and arrived at an unglamorous answer: people could not be bothered, or they typed in rubbish to make the requirement go away. Her response was to write digital adoption into the sales management standards — a defined picture of good practice for a rep and for a first-line manager — on the reasoning that the first-line manager is where execution either happens or stops. If that layer treats the tool as optional, everyone below it will too.

A healthcare CEO put the discipline on it that the room kept returning to: much of the adoption push had been adoption for its own sake. The questions worth asking first are which problem is being solved and whether it is one the business cares about.

The AI that earned its budget

The sharpest AI account came from a regional VP at a global taste and nutrition business, and it opened with a note of impatience. Everyone talks a good game about their investment; measured at the point of execution, he was disappointed in what his own industry was producing. His read was that the wrong people have the tools and few of them are talking to the front line.

1. 1

   The data was already there, unread

   Three to four hundred sellers in his region, filing roughly four visit reports each per week. Tens of thousands of data points a year, fragmented across the organization, with the consolidation work left undone. In B2B there is no point-of-sale data to fall back on, so the visit report is the asset.

2. 2

   Consolidate first, then synthesize

   The call reports each contained market insights — segment clusters, market trends, customer trends, innovation trends. Stored in one place and put through an off-the-shelf model, the fragments became proprietary insight the company could take back to its customers at scale. Competitors have no route to it, because they do not have the visits.

3. 3

   Collapse the innovation cycle

   Food and beverage concepts reach consumers eighteen to twenty-four months after the work starts. The question he put to his team was whether ninety percent of the time between an idea and a tangible concept could be removed. Using a tool that scrapes the web against industry keywords, paired with market visits and a customer brainstorm, a session now runs from an idea generated on a market visit to a pictured concept with a full recipe — inside the same meeting.

A digital and commercial excellence VP at a large technology distributor described the arc his organization had been through, from an initial belief that AI would lift the company's valuation on its own to a more useful position. He had made himself an AI champion for sales growth on the explicit basis that the technology is somebody else's problem: bring the business outcome, then earn adoption on the use case.

Guardrails, and the case where the answer inverts

A general manager recently out of specialty chemicals named the spread he had watched inside one organization: people putting everything they know into a model, people using it for small efficiencies, and people staying away from it. His concern was the middle of that — if the company offers no guidance at all, the behavior fills the vacuum on its own.

The answers split by what kind of data is involved, which is the part worth keeping.

• Where the data is owned and contained, the problem inverts. The taste and nutrition VP wanted more going in, not less — the visit reports sit behind the company's own firewall, and the risk he was managing was sellers filing thin reports rather than sensitive ones.
• Where the data is regulated, gatekeeping is the job. A healthcare CEO working with protected health information described processes, training and named gatekeepers to keep it out of publicly accessible models.
• Company-provided beats company-permitted. A commercial excellence lead in healthcare made the point that consumer tools are available to everyone anyway; what changed the use cases was the company subscribing and providing a safe environment where internal information could go in.

The same participant had found the value in narrow agents rather than general chat — one for email, one for meeting preparation — each doing a defined task rather than being asked to be useful in general.

The counterexample came from the commercial director for a water utility serving the Caribbean, and it was a useful corrective to a room full of enthusiasm. His customers are governments, and they want AI nowhere near the drinking water.

A strategic pivot is a commercial rebuild

The most structural account of the year came from a global commercial excellence director in building materials, whose company had changed ownership and, with it, its objective — out of margin optimization and into revenue growth and cash flow. He had run this kind of role for a decade across several companies and was candid that he had understated the depth of it.

1. 1

   New KPIs

   The existing measures had been built to guide margin decisions, and left in place they keep producing the decisions they were designed for.

2. 2

   New analysis and new segmentation

   Understanding the market at a different level, because growth pockets sit outside the zone of comfort and the old segmentation had no reason to find them.

3. 3

   A different risk profile

   Going to places the company had avoided, which means accepting a class of risk the margin-first model had been screening out.

4. 4

   New rules of engagement

   Deals negotiated differently, price levels the company had declined before, and a story for the customer explaining why — without contaminating the rest of the portfolio. That demanded a more differentiated value proposition and pricing strategy than had ever been needed.

5. 5

   A rebuilt sales incentive plan

   Rethought completely rather than adjusted. Motivation was the piece he came back to, and a plan written to reward margin will keep producing margin behavior whatever the strategy says.

Over-engineering, named repeatedly

A global head of transformation at a logistics company described a KPI standardization program running for the same reason. Hundreds of measures, many of them bespoke to individual customers and most of them operational enough to make a senior conversation hard to have. The standardization question was which ones contribute to value for the customer.

A commercial excellence director in chemicals took that further, and got the strongest reaction of the session for it. Companies want to fix pricing, planning, pipeline management, CRM and portfolio simultaneously. It rarely works, for two reasons: the resources are not there, and each change lands on something another change is already moving.

A sales excellence head in flow control named the pattern behind it — corporate life adds layers on layers, and the word excellence gets heard as a promise of a magic wand.

His conclusion was to give the time back: stop the over-engineering, spend it with the front line, and be willing to name the waste plainly enough that senior management can act on it.

Tariffs: cut the noise, then read the contracts

The most operationally difficult account came from the Caribbean, where a regional commercial director covers fourteen countries and more than thirty governments across Dutch, French, British and US affiliations. Tariff changes arrived weekly and sometimes daily, prices were updated daily, and the deals still had to close.

What his team settled on was to stop trying to model the whole picture and instead ask each government directly where they wanted goods sourced from and how quickly they needed them. Some traded speed for price; others reversed it, accepting delay to keep shipments away from routes that would multiply the cost. It reads like going backwards thirty years, and it was the thing that worked.

A commercial excellence lead in paints and coatings offered the structural version of the same lesson. His company had centralized operations for synergy before the pandemic and decentralized afterwards to sit closer to its markets. When the tariffs arrived and a manufacturing route from Vietnam into the US collapsed, local-for-local turned out to have been partial protection bought for other reasons.

A departing chief commercial officer in medical technology gave the room its planning frame. Large industrials had already stated their worst-case exposure publicly — half a billion dollars off the bottom line, in the examples he cited — and the value of a number like that is that it ends the argument and starts the plan.

1. 1

   Map the flows

   Financial flows and goods flows, both. The right answer depends on the business model: a flow product with a short order-to-cash cycle behaves differently from a backlog product priced now and invoiced six or twelve months later.

2. 2

   Stand up an office of tariffs

   Finance, project management, sales and legal in one team. Sales because someone has to have the conversation with the customer, legal because the answer is usually in the contract.

3. 3

   Read the escalation clauses

   Where a contract was written without one, the exposure sits with you. At some stage the customer takes a share of the hit, and which stage that is was decided when the contract was drafted.

4. 4

   Set the mitigation milestone

   Short-term mitigation may be out of reach. What matters is a dated point by which it is fully mitigated and off the bottom line.

The work tariffs force — tracing supply chains, mapping goods and financial flows, centralizing what had been handled locally — is analysis most companies had been postponing, and the visibility outlasts the tariff that prompted it. One further caution came from the building materials director, learned the hard way: do not assume competitors and other players in the chain will behave the way you would. That assumption tends to be wrong.

The golden bridge

The story that closed the session came from a global transformation lead in supply chain at an appliance manufacturer, who opened with the line from Ted Lasso that had kept him going through the year.

His company is organized in regions — Europe, Asia, North America, Latin America — each of which had operated, in his description, as its own kingdom. Tactical planning had been implemented region by region: the same ERP, configured differently everywhere. Bringing that together had taken three years. The next layer down, sales and operations execution, was about to repeat the pattern, with each region preparing to run its own tender.

1. 1

   Find the person who can build the bridge

   He assumed the IT VP would be the sponsor and was wrong — that VP was settled in the existing habits. The person who mattered turned out to sit in purchasing, because the tenders had to pass through there. Two or three more stakeholders followed.

2. 2

   Co-create the governance rather than issue it

   The program had no governance board. He built one with the people who would sit on it, which made them part of the thing rather than subject to it.

3. 3

   Wait for the small crisis

   Progress was slower than expected until an investment request ran into approval trouble. That supplied the urgency, and the plan was ready before the urgency arrived.

The outcome was one tender and one implementation across all regions. The patience is the part that is hard to copy: for most of the year the work looked like nothing was happening.

Global standard, regional agility

The technology distributor's VP put a limit on how far that generalizes. With eighteen markets in Europe at different levels of technology maturity, waiting for global harmonization on each of those decisions would have meant waiting years. A global strategy and direction with regional flexibility beneath it was, in his organization, the differentiating choice rather than the compromise.

The supply chain lead agreed and drew the line where the data is: make the core data available to everyone, then allow best-of-breed additions and local initiatives on top. He was open that the boundary is unresolved — how much dashboarding should be standard against self-service, and how a locally built dashboard gets promoted for wider use, are questions he does not have clean answers to.

The logistics transformation head added the economic argument for pulling local work upward. The local organizations sit closest to the customer and generate the ideas; a company can afford to invest once, rather than seventeen times across seventeen markets. That makes capturing local innovation and delivering it back out a funding decision as much as a community one.

Training is not enablement

The lesson that drew unanimous agreement, and the one most of the room said it was taking into 2026, came from the logistics transformation head.

• Role modeling — leading from the front, demonstrating the practice rather than sponsoring it.
• Staying with people through the first application, where the attempt either sticks or gets abandoned.
• Understanding why someone would be motivated to accept the new tool or process, rather than assuming the mandate covers it.
• Building the confidence to apply the tools and frameworks without supervision.

The supply chain lead confirmed it from the other direction, and put a cost on the alternative. Traditional change management — the stakeholder maps, the formalization, the PowerPoint communications program — wasted a great deal of money in his experience. What worked was closer to a lean discipline: guided discovery, going to where the work happens, and coaching.

One idea for 2026

A former P&L owner in chemicals closed the discussion by connecting the AI conversation back to a problem the room has lived with for twenty years. Sales force effectiveness studies used to show sellers spending fifteen to seventeen percent of their time in front of customers, with the rest going to admin, reporting and internal writing. Absorbing that work is a good use of the technology and largely settled.

His unsettled idea was better: use it to help sellers ask better questions in the room. Asking is the thing sellers most often skip, because they talk and overwhelm the customer instead — and it is a coaching problem that has resisted the training programs aimed at it.