New Math, Old Muscle: The Midmarket Edge 

For most of the last decade, pricing sophistication in industrials belonged to the biggest players. Large companies built pricing science teams, licensed enterprise software, and ran multiyear transformation programs. Midmarket industrials, the businesses that make up most PE portfolios, usually couldn’t justify that cost and made do with what they had. AI is changing that equation faster than most midmarket leaders realize, and for once the shift favors them.

The pressure to act is already there. At many midmarket industrials, materials, labor, and freight have outrun price increases, and multiple expansion is no longer around to cover the gap. Pricing is the most direct answer. It moves EBITDA without new capital or new customers, and the data it needs already exists in the ERP. On a $300 million business, one point of realized price is $3 million of EBITDA before anything else changes. It doesn’t need a large AI budget. It needs one well-chosen bet and the discipline to see it through.

The math used to be the moat. Industrial pricing has always been a judgment problem. Rules were too rigid for the deals that mattered most, models were too narrow to see the whole deal, and experts were too few to be on every quote. Large companies could at least put more people on the problem. Midmarket companies settled for averages: price lists, discount bands, and an annual review. In the last year, AI learned to reason through a multi-step analysis, hold the playbook, contracts, and account history in view at once, and explain its recommendation in plain language. A question that used to cost an analyst a day, why margin on a product line fell two points and what to do about it, now gets a specific answer in minutes. A midmarket company with one pricing analyst, or none, can now have guidance on every deal, from first ask to signed quote.

That levels the math. It doesn’t level the outcome. In our experience, the algorithm is the smallest part of the value. Most of it lives in the people, processes, and workflows around the model. That is what we call the old muscle, and it is where the midmarket has both a real edge and a real gap.

The edge is proximity. In a midsize industrial, the CEO, the head of sales, and the pricing lead are often in the same weekly meeting. A new approval threshold can be set on Monday and be in reps’ hands by Friday. The sales team is small enough to coach one rep at a time, and leaders tend to own outcomes personally rather than handing them to a steering committee. A pricing change that has to work its way through regions, business units, and IT governance at a large company can often move in a single quarter at a midmarket one.

The gap is capacity. Midmarket companies tend to have thin pricing benches, messy data, and no one with the time to own the change. AI won’t fix that data. It will expose it. And someone still has to walk the rep through the trade-off, reset the guardrail, and check whether the order landed at the quoted price. That is why so many AI efforts stall between a promising pilot and a changed process. The speed advantage is real, but only if someone is there to use it.

That is also why we think the build-versus-buy debate is usually framed wrong. For a midmarket industrial, the answer is both, at different layers. Buy the math, because building pricing models from scratch is slow and rarely a source of advantage. Build the muscle, and if the bench is thin, borrow it from people who have done it before. That is where the value is captured, and it does not come in the box with any tool.

The clock matters, too. Many midmarket industrials sell into customers far larger than they are, and those customers’ procurement teams are building AI right now to compare suppliers, push for discounts, and hold sellers to their terms. That sets up a negotiation between a buyer with a large-cap AI budget and a seller with a midmarket one. The seller’s best defense is not a bigger budget. It is knowing its own margin better than the buyer does, on every deal.