Retention is treated as an operational matter to be improved after completion. It is more usefully treated as a valuation input to be established before it. Churn that emerges in year one was almost never created in year one — it was already present in the base, visible in the data, and simply not asked about. That makes it a diligence failure rather than a management one.
A one point movement in gross retention compounds. Over a five year hold it can move exit revenue by more than most pricing initiatives, which is why retention deserves the diligence attention that pricing usually gets.
Reported retention is a lagging indicator of a decision already made
By the time a customer churns, the decision is old. In most B2B businesses the disengagement that produces a non-renewal begins somewhere between six and eighteen months earlier: a champion leaves, a use case is not extended, usage plateaus, a support experience goes badly, a competitor is evaluated quietly.
This means reported retention describes decisions taken well before the period being measured. A business showing 94% gross retention today is showing you the outcome of relationships as they stood a year or more ago — under the previous plan, the previous product and often the previous team. It is genuinely useful evidence, and it is evidence about the past.
The consequence for diligence is direct. If you want to know what retention will be under your ownership, historic retention is the wrong instrument. You need the leading indicators, which are almost always available and almost never requested.
The leading indicators, in order of usefulness
| Signal | Where to find it | What it predicts |
|---|---|---|
| Usage or consumption trend by account | Product telemetry, billing records, licence activation | The single strongest predictor available. Accounts declining in usage for two consecutive quarters churn at materially higher rates regardless of what the relationship feels like. |
| Champion turnover | CRM contact records, or simply asking who the main contact was 18 months ago | A changed primary contact resets the relationship. If nobody rebuilt it, the renewal is exposed. |
| Support and escalation history | Ticket system, severity and resolution time by account | Concentrated unresolved escalations in large accounts are a renewal risk that never appears in a commercial deck. |
| Breadth of deployment | Seats or modules purchased versus active, by account | Single-team deployments in large organisations are fragile. Multi-team deployments rarely churn wholesale. |
| Renewal lead time | Renewal calendar with owner and status | Renewals being worked in the final fortnight indicate no renewal motion. Retention is currently a function of inertia. |
| Price increase history | Realised price by account over time | Accounts that have absorbed increases without incident are durable. Accounts never tested are unknown, not safe. |
The first row is the one worth fighting for. Where usage data exists it usually settles the question faster than everything else combined, and it is the item most frequently missing from a commercial data request.
Recorded churn reasons are usually wrong
Ask any business why customers leave and a large share of the recorded reasons will say price. Price is what customers say, because it is the least confrontational answer available and it ends the conversation politely. It is rarely the operative cause.
Customers who are deriving value absorb price increases routinely. Customers who are not, leave and cite cost. The recorded reason therefore tends to record the exit conversation rather than the decision, and a business that logs it uncritically has not asked the second question.
Two practical corrections during diligence:
- Speak to customers who left. Two or three conversations with churned accounts are worth more than the entire churn reason field. They are also easier to arrange than people expect, because former customers have no reason to manage you.
- Cross-check reason against usage. If accounts citing price were also declining in usage beforehand, the cause was value realisation. If they were fully engaged and still left on price, you may have a genuine pricing problem — a much rarer and more tractable finding.
Net retention hides two different businesses
Net revenue retention above 100% is reported as a single achievement. It is usually the sum of two independent effects that need separating, because they have opposite implications for a buyer.
Expansion can come from growth in value delivered — more teams, more use cases, more of the product genuinely adopted. Or it can come from contractual escalators and price rises applied to a static base. Both lift net retention. Only the first indicates a product customers want more of.
The test is to compute net retention twice: once as reported, and once with all price effects held flat. The gap between the two figures tells you how much of the growth story is adoption and how much is indexation. A business at 112% net retention that falls to 99% on constant pricing is a very different asset from one that falls to 108%, and no reported metric distinguishes them.
Gross retention deserves equal billing for the same reason. It cannot be flattered by expansion, which makes it the cleaner read on whether customers actually stay.
What to do with the finding
Retention findings resolve into three broad outcomes, and each has a different consequence for the deal rather than merely for the plan.
- Retention is real and managed. Named owners, early warning, renewals worked months ahead, usage healthy. This supports the multiple and should be said explicitly, because it is a genuine quality signal that often goes unremarked.
- Retention is real but unmanaged. Customers stay because switching is hard, not because anyone is working at it. Durable for now, fragile to a competitive shift, and cheap to fix. This is an opportunity: building a renewal motion is one of the highest-return, lowest-cost workstreams available post-close.
- Retention is deteriorating and the base is exposed. Usage declining, renewals concentrated in the next eighteen months, escalations unresolved. This is a pricing matter, not a plan matter. The model needs rebuilding before the offer, not fixing after it.
The distinction between the second and third outcomes is what the leading indicators buy you, and it is not visible in the reported retention number that most processes stop at.
Frequently asked questions
What is the difference between gross and net revenue retention?
Gross retention measures how much of the existing revenue base is kept, and cannot exceed 100%. Net retention includes expansion within that base and can exceed it. Gross retention is the cleaner read on whether customers stay, because net retention can be lifted by price increases applied to a shrinking base.
How far back should churn analysis go?
Three years of customer-level revenue by period, with start dates, segments and churn dates. Less than that makes it hard to distinguish a genuine trend from a single bad cohort or a discontinued acquisition channel.
Why interview customers who have already left?
Because they have no relationship to protect and will be more precise than current customers about what went wrong. Two or three such conversations typically produce more diagnostic value than an entire field of recorded churn reasons.
Should churn affect the price or the plan?
Both, depending on the finding. Retention that is unmanaged but currently holding is usually a plan item and a cheap one. Retention that is actively deteriorating with an exposed renewal calendar is a valuation input and should be reflected before the offer.
Ready to aim higher?
If your model depends on the customer base holding, that assumption is testable before completion rather than after it. We assess retention alongside the rest of the commercial engine and deliver a prioritised plan.