Verifying the retention story behind a growth-stage investment
A headline retention claim was rebuilt from raw billing data, tested against customer evidence, and translated into the growth assumptions embedded in valuation before a nearly signed transaction.
- Client profile
- Angel syndicate
- Target profile
- Growth-stage company
- Decision window
- Three weeks before a seven-figure allocation
- Resulting decision
- Renegotiate rather than walk
The decision behind the brief
The lead investor did not need another opinion about whether retention looked impressive. The decision required an independent reconstruction of what the metric meant and what the valuation assumed.
An angel syndicate had a term sheet nearly signed for a growth-stage company. The company's deck claimed best-in-class net revenue retention, and the lead wanted independent verification on a three-week clock before wiring a seven-figure allocation.
The decision was consequential and time-bound. Accepting the summary metric without testing its construction could overstate business quality. Treating any discrepancy as a reason to walk could also discard a transaction whose terms might be corrected. The work had to distinguish between a weak company, a weak metric definition, and an operating issue that could be monitored.
Constraints that shaped the work
- The deal clock was three weeks. Analysis had to focus on evidence capable of changing the transaction decision.
- The headline metric was definition-sensitive. The published case states that downgrades were excluded from the reported figure.
- One cohort could distort the aggregate. An anomalous enterprise cohort materially supported the headline result.
- Customer evidence needed disconfirmation. References had to test the retention narrative rather than merely collect supportive quotes.
- Commercial evidence had to connect to price. Correcting retention mattered only if the impact on the valuation assumptions could be made explicit.
Evidence used in the engagement
The published case identifies the following evidence. It does not disclose the sector, billing-history span, cohort count, original or corrected retention rates, full reference-call count, or valuation-model assumptions.
- Raw billing exports were used to rebuild cohort retention rather than accepting the summary tab.
- Logo and revenue retention were separated so customer loss, expansion, contraction, and concentration were not collapsed into one figure.
- An anomalous enterprise cohort and excluded downgrades were identified as material features of the headline claim.
- Reference calls designed for disconfirmation included two churned accounts the company volunteered reluctantly.
- A reverse-DCF framing connected the corrected retention evidence to the operating performance required by the proposed valuation.
Analytical approach
Rebuild the metric from transaction-level evidence
Cohort retention was reconstructed from raw billing exports rather than the summary tab. This made the treatment of starting revenue, expansion, contraction, churn, and cohort membership inspectable.
General method note: A retention metric should be rebuilt from a documented cohort definition and reconciliation rules. Small changes in exclusions, period boundaries, and account mapping can materially change the result.
Separate customer survival from revenue movement
The work separated logo retention from revenue retention. It also found that the headline figure excluded downgrades, which meant the deck's metric did not capture all contraction within the retained base.
General method note: Logo retention answers whether customers remain. Gross revenue retention adds contraction. Net revenue retention also includes expansion. No one metric can diagnose all three mechanisms by itself.
Interrogate the aggregate
The headline result leaned on one anomalous enterprise cohort. Looking at cohorts separately prevented a large or unusual group from being mistaken for a broad, repeatable pattern.
General method note: Aggregate retention can hide concentration, cohort-age effects, segment mix, and survival bias. The first task is not to discard an outlier automatically, but to explain why it differs and whether that mechanism can recur.
Design references to challenge the favored explanation
Reference calls were structured for disconfirmation and included two churned accounts. Those calls surfaced a consistent onboarding gap rather than merely confirming general customer satisfaction.
General method note: Customer references are more decision-useful when the sample includes losses and weak-fit accounts, questions test competing explanations, and findings are graded by source quality rather than counted as votes.
Translate operating evidence into valuation expectations
A reverse-DCF framing showed that the proposed valuation required retention the corrected cohorts did not support. This connected a measurement issue to the transaction terms rather than leaving it as an isolated analytical observation.
General method note: Reverse DCF starts with price and solves for the performance needed to justify it. It does not establish intrinsic value by itself; it makes the embedded operating expectations visible for comparison with the evidence.
How the evidence changed the transaction frame
| Claim or question | Published finding | Decision implication |
|---|---|---|
| Was reported NRR representative? | The headline leaned on one anomalous enterprise cohort and excluded downgrades. | Use corrected cohort evidence rather than the deck metric as the underwriting baseline. |
| What might explain the weaker pattern? | Disconfirming reference calls surfaced a consistent onboarding gap. | Treat onboarding as an operating issue to monitor, not merely a narrative explanation. |
| Did the proposed valuation fit the evidence? | Reverse-DCF framing required retention the corrected cohorts did not support. | Revisit transaction terms rather than accept the existing sheet unchanged. |
| Walk or renegotiate? | The issue was translated into revised economics and a monitorable operating metric. | The syndicate renegotiated and added an onboarding-metrics information right. |
Resulting decision and monitoring logic
The syndicate renegotiated rather than walked. Terms were restructured at a valuation approximately 25% below the original sheet, and an onboarding-metrics information right was added.
The corrected retention model became the syndicate's monitoring baseline. This preserved the analytical definition used during diligence and connected the identified onboarding concern to information available after signing.
Observed outcome
The transaction proceeded on revised terms with a clearer operating baseline
The company subsequently hit its revised plan. The corrected retention model became the syndicate's monitoring baseline, and the published case reports that the engagement paid for itself several times over at signing through the change in terms.
Jagdeep Ventures provides independent research and analysis, not investment advice. Investment decisions remain with the client and their licensed advisers. See the research disclosures.
Limitations and attribution
The facts are anonymized, lightly generalized, and rounded. The public case does not provide the target sector, cohort definitions, billing-history span, original or corrected retention rates, complete reference-call sample, reverse-DCF assumptions, exact transaction terms, or the period over which the revised plan was assessed.
The lower valuation and information right were transaction outcomes after the diligence. The public record does not establish that the analysis alone caused the company's later performance, and a good decision process should be judged separately from an outcome that can still be influenced by many external factors.
Transferable lessons
- Rebuild decision-critical metrics from raw records and document every inclusion rule.
- Inspect cohorts and concentration before treating an aggregate as representative.
- Use customer references to test competing explanations, including churned accounts when access permits.
- Translate a diligence finding into price, terms, and a post-close signpost rather than stopping at a red flag.
Need an independent read before a transaction?
Describe the decision, the evidence available, and the deadline. We will tell you which commercial questions can be tested in time and where uncertainty will remain.