Composite case
This autopsy is a composite. It reflects patterns SLAM sees repeatedly across commercial due diligence, assembled from multiple engagements and observable market dynamics. It is not any single company, and any resemblance to a specific business is coincidental.
The Setup
The Business
Usage-priced SaaS. Data infrastructure tooling sold to mid-market and enterprise engineering teams.
The Numbers
£22M ARR with reported net revenue retention of 125%.
The Deal
Acquired by a PE firm at a premium multiple, priced for durable net expansion.
The Thesis
The base compounds on its own. Buy the expansion engine and let land-and-expand carry the return.
The whole case turned on one number. At 125% net revenue retention, the existing customer base grows revenue every year without a single new logo, and a buyer will pay a premium for that compounding. The model assumed the expansion continued at something like the historical rate. Everything else in the thesis was secondary to whether that assumption held. It did not, and the reasons it did not were all present at close.
What Diligence Saw
The retention story was compelling, and every headline metric reinforced it.
125%
Net revenue retention
Read as elite
96%
Logo retention
Read as sticky
Rising
Average contract value, expanding cohort by cohort
Read as durable
Land & expand
A clean, repeatable expansion motion on paper
Read as a moat
Net revenue retention is the metric a SaaS multiple is most often built on, and this one was excellent. But NRR is an aggregate, and an aggregate can be excellent for reasons that do not repeat. The figure was strong. Its two most important properties, durability and organic growth, were both partly illusory.
The Unravelling
Deal closes. Premium paid for durable expansion.
The model assumes net revenue retention holds near 125% through the hold period.
The first renewal cohort pushes back.
Several of the largest, fastest-growing accounts have crossed a usage-pricing tier. The step-change in price triggers procurement scrutiny that a smooth curve never would have.
The "expansion" is examined for the first time.
A meaningful share of last year's expansion turns out to be customers consolidating budget from a cancelled adjacent tool. It was recorded identically to organic growth. It is a one-time reallocation.
A silent cohort surfaces in the usage data.
A set of accounts has been quietly declining in engagement for months, with no support tickets and no complaints. Still inside contract, still counted in NRR, already gone in everything but paperwork.
The tier-boundary renewals churn or renegotiate down.
The accounts the model expected to expand most are the ones the pricing cliff punishes hardest. The finite pool of consolidatable budget is running out.
NRR restates from 125% to around 103%.
The durable, compounding net expansion the premium was paid for was neither durable nor, in meaningful part, expansion. The engine was a finite reallocation sitting on a quietly churning base.
The Diagnosis
A single headline retention figure concealed three distinct problems, each attacking a different property of that figure. One inflated it, one made its healthiest-looking accounts fragile, and one hid a loss the metric had not yet registered.
A significant share of reported expansion was replacement spend, budget moved into the product from a cancelled adjacent tool, recorded identically to genuine usage growth. Replacement spend is a one-time reallocation. It does not repeat, and once the readily consolidatable budget across the base is exhausted, the expansion rate falls whatever the sales motion does.
Usage-based pricing meant the most successful customers, the ones whose growth the model most relied on, hit a step-change price increase at renewal. A buyer's finance function reacts to the rate of change, not the absolute level, so a sudden jump triggered procurement scrutiny and churn precisely where the thesis expected the most expansion.
A cohort of accounts had already disengaged. Usage was declining, executive contact had gone quiet, and because nothing was overtly wrong, no support ticket or escalation ever flagged them. They were still inside their contracts and still in the NRR denominator, six months from a churn the metric could not yet see.
Why It Compounded
The three combine to make one number lie in three directions at once. The mirage inflated the expansion numerator with spend that would not recur. The renewal cliff turned the healthiest-looking accounts, the expanders, into the churn risk. And silent resignation hid the denominator loss that was already coming. Read together, the 125% was a finite budget reallocation stacked on top of a base that was quietly leaving. It looked like compounding growth. It was the opposite, disguised by an aggregate.
The Cost
125% → 103%
Net revenue retention, restated inside Year 1
Finite
The expansion pool, once consolidatable budget ran out
Premium
Paid for durability the base did not have
Reforecast
The compounding return thesis, revised down
What a Behavioural Diligence Would Have Caught
None of the three is hard to test for. Each requires only that the diligence interrogate the retention figure instead of accepting the aggregate.
Source-of-Budget Test
Interviewing a sample of expansion accounts about where the incremental budget came from would have separated organic growth from one-time replacement spend, and revealed how much of the expansion was reallocation.
Tier-Boundary Test
Plotting renewal outcomes against each account's proximity to a pricing tier would have exposed the cliff: churn and heavy discounting clustering exactly where customers crossed a threshold.
Usage-Decay Test
Reconstructing NRR from raw product usage rather than the renewal model would have surfaced the silently resigning cohort months before it showed up as churn.
The retention figure was real as a number and misleading as a signal. A diligence that took it apart, rather than taking it as given, would have priced the base for what it was.