Revenue up 140%, and each new cohort worse than the last

The headline growth was genuine. The cohort table showed the company was refilling a bucket that was emptying faster every quarter.

This is an anonymised composite, not a report about a named company or person. The situation and numbers are typical rather than reported.

Where they were

A B2B software company with 2,400,000 of ARR, up 140% year on year, selling to small professional services firms. The deck led with the growth rate and a logo wall. On request, the company provided retention by signup cohort: customers who joined two years earlier retained 71% of revenue after twelve months; the cohort from the following year retained 54% over the same period. Net revenue retention had fallen from 112% to 94%. The round was 6,000,000 at a 24,000,000 post-money valuation.

The problem

Two facts pointed in opposite directions. Revenue growth was accelerating, which suggested demand; cohort retention was deteriorating, which suggested the product was not delivering lasting value to the customers being added. Only one of those facts would still matter in three years.

The decision

Invest on the growth, investigate the retention decline further before deciding, or decline on the cohort evidence.

What was on the table

• Invest at 24,000,000
  For: 140% growth with 2,400,000 of ARR is genuinely hard to achieve, and retention problems are sometimes fixable with product work and better customer selection.
  Against: Falling cohort retention with net revenue retention below 100% means existing customers shrink the revenue base, so all growth must come from new sales, which gets harder and more expensive.
• Spend two more weeks on retention diligence
  For: Speaking to churned customers would show whether the cause was fixable - the wrong customers being sold to - or structural, such as the product solving a one-off problem.
  Against: The round was competitive and closing; the investor risked doing the work and losing the allocation anyway.
• Decline on the cohort evidence
  For: A deteriorating cohort trend is among the few metrics that is hard to reverse and expensive to be wrong about, and the price assumed the growth would continue.
  Against: Retention can improve substantially when a company narrows its customer focus, and declining meant giving up on a fast-growing company for a problem that might be solvable.

What they chose

The investor spent four days on retention diligence rather than two weeks, speaking to six churned customers. Five described the product as useful for a one-time migration task and unnecessary afterwards. That answer made the problem structural rather than a sales-targeting error, and the investor declined, recording the churn interviews as the deciding evidence.

What happened

Growth continued for a further eleven months, reaching 3,600,000 of ARR, before new sales slowed and the shrinking base became visible in the aggregate. ARR fell to 2,900,000 over the next year. The company raised a down round at a 14,000,000 post-money valuation with a 1.5x participating preference, which left earlier holders substantially diluted. The product remained genuinely useful for the task it solved; the task simply did not recur.

What transfers

• Cohort retention is the metric growth cannot disguise, and it is worth the awkwardness of asking for it by signup month rather than as a blended figure.
• Net revenue retention below 100% means the existing base shrinks, so growth depends entirely on new sales at rising cost.
• A handful of conversations with churned customers often resolves in days what financial analysis cannot resolve at all: whether a retention problem is structural or a selection error.
• Speed of diligence can be chosen. Four focused days on the single deciding question beat two weeks of broad work that would have missed the closing date.

More case studies