There is a question circulating in M&A diligence rooms in 2026 that barely existed three years ago. Not "what is your NRR?" — every founder who has spoken to a buyer knows that question. The new question is this: "Walk me through how it is generated."
For founders who have a clear answer, the conversation moves forward. For those who state a percentage and wait, a risk flag goes into the model and the multiple starts to adjust.
Two businesses I am aware of from recent advisory work illustrate why this matters. Both operated in the same UK SaaS sector. Both had £10M ARR, 27% annual growth, and 115% net revenue retention. Both ran structured sale processes with the same broad universe of strategic and financial buyers.
One business closed at £52M — 5.2x ARR. The other closed at £78M — 7.8x ARR.
The £26M difference had nothing to do with the NRR figure. It had everything to do with the mechanism behind it.
Business A — Seat Expansion
Business B — Multi-Mechanism
The seat expansion story
The first business had built its NRR on a mechanism that worked reliably for five years. When clients grew their teams, licences grew with them. The product was seat-based, and as customers hired, they provisioned more users. This produced NRR of 110% or above every quarter without active sales effort — it was organic, automatic, and attractive.
The buyer's diligence team ran a cohort-level analysis of seats per customer over time. What they found was a trend that had quietly shifted over the prior eight months. Average seats per customer had declined from 38 to 31. Customers were not churning. They were consolidating. Functions that previously required eight or ten human users were being handled by two or three, with AI tools filling the gap.
The trailing 12-month NRR figure was still 115%. The forward NRR assumption the buyer was willing to model was considerably lower.
The repricing logic: When a buyer identifies a structural trend behind a metric, they do not simply accept the trailing number. They project forward. If seat count per customer declined 18% over eight months, the buyer's model assumed continued pressure under accelerated AI adoption. The multiple adjustment reflected this forward risk, not the historical performance.
The founder's NRR was real. The problem was that the mechanism generating it was no longer neutral to AI-driven changes in how customers staff their teams. Buyers had seen similar patterns in other portfolios. They priced for the risk.
The multi-mechanism story
The second business had built its retention differently, though the total NRR figure looked identical. When the buyer asked the mechanism question, the founder spent eleven minutes on the answer.
The business had three product modules. In the first year of a new customer relationship, clients typically purchased one. By Year 3, 68% of the base was using at least two. This multi-product adoption had not happened accidentally — it was the result of a deliberate customer success motion launched three years earlier.
Alongside the product expansion, the business had a history of annual price increases. Between 7% and 9% per year, applied systematically across the base, contractually embedded with CPI-linked escalator clauses. These clauses had never been contested in six years of use.
Beyond these two primary mechanisms, a third — cross-sell of a professional services tier — contributed a smaller proportion. A fourth, international upsell as certain customers expanded into new territories, contributed a further slice.
Four mechanisms. Each independently defensible. Each backed by cohort-level data the founder presented proactively, without being asked.
The buyer's model had no single point of failure to probe. The 115% was distributed across drivers that carried different risk profiles. Multi-product adoption reduces churn. Contractual price escalators are not exposed to headcount trends. The combined story held under scrutiny.
Why 2026 changed the question
This distinction — between NRR as a number and NRR as a mechanism — matters more now than it did in 2022 or 2023. Two things have changed.
The first is that buyers, particularly strategic acquirers, are building AI into their diligence workflows. They process cohort data faster than any human team could three years ago. The seat count decline described above — from 38 to 31 per customer — was identified in under 24 hours of diligence.
The second is that the AI adoption curve in enterprise customers has made seat-based expansion a less reliable forward assumption across a wide range of software categories. For a seat-based SaaS business, the question "is this customer's licence count likely to grow or shrink over the next 24 months?" is now part of standard underwriting.
Market context: In the 2025–2026 European lower mid-market, businesses with multi-mechanism NRR — price escalators plus multi-product penetration plus geographic expansion — have been exiting at 6.5x–9x ARR. Seat-based businesses with otherwise comparable metrics have seen multiples between 4x and 6x, with the gap driven primarily by forward NRR assumptions in the buyer's model.
What buyers are actually asking
The mechanism question has become standard in informed diligence rooms. What proportion of your NRR above 100% comes from seat or user expansion? What is your price increase history, and is it contractually supported? What is your multi-product adoption rate, and what is the trend? What customer segments drive each mechanism?
These questions are not a surprise if a founder has prepared for them. They are a significant problem if encountered for the first time in a buyer meeting.
Preparing the NRR narrative
The difference between the two founders in this example was not the quality of the businesses they had built. It was the quality of the data they had prepared and the narrative they had constructed around it.
The founder who answered in eleven minutes had done this work in advance. She had a mechanism-level breakdown of her trailing 24-month NRR, showing the contribution of each driver by quarter. She had a pricing history document that showed every price increase applied over six years, the contractual basis for each, and the customer acceptance rate. She had a cohort chart showing multi-product adoption by customer vintage.
None of this data was fabricated or spun. It was the same underlying business as her competitor in the process. The difference was that she had organised the data to tell the mechanism story clearly, before any buyer asked.
The founder who answered in two minutes was not hiding anything. He simply had not prepared the decomposition. He knew his NRR was 115% and trusted that the number would carry the conversation. In the market of 2022, it might have. In 2026, it did not.
The CEO Tip
Before any buyer conversation, decompose your NRR by mechanism. Seat expansion, price increases, upsell to higher tiers, cross-sell of additional modules, and geographic expansion within existing accounts should be reported as separate line items, each with a trailing 12-month contribution figure.
If one mechanism is dominant and that mechanism carries AI exposure — typically seat-based expansion in functions where AI is substituting for headcount — you have time to diversify the story before entering a process. The mechanism question is answerable. But the time to build the answer is before the buyer asks it, not during the diligence call.
The £26M gap between these two businesses was not a market anomaly. It was the market correctly pricing the difference between durable and fragile retention. Buyers are getting better at identifying which is which. Founders need to be equally clear about what they have built.