NRR Falls from 118% to 104% with Seven CSMs and a Green-Light Health Score
Aperture Learning sells to mid-market and SMB customers. Net revenue retention fell from 118% to 104% over four quarters, with gross revenue retention at 91%. Logo churn is concentrated in accounts under $15k ACV, which are 62% of logos but 18% of ARR and churn at 26% annually. The $50k-plus segment retains at 95% and expands at 131%. Seven CSMs each cover about 180 accounts, plus a shared support queue. The health score is built from product logins and ticket volume. A post-mortem of 46 churned accounts found 19 had green scores 60 days before churn, and 12 had never completed onboarding; median time to first value is 74 days against a 30-day target. The CRO wants to hire four more CSMs at $95k fully loaded. The CFO wants to cut the SMB segment. Only 8% of accounts had an executive business review in the past year.
1. Decide where the next dollar of customer success capacity goes. Show the math.
Read the worked answer
Work the segment economics before adding or cutting anything.
If total ARR is roughly $40M, the under-$15k segment is about $7.2M with 26% annual churn, so roughly $1.9M of ARR is at risk there each year. The $50k-plus segment retains at 95% and expands at 131%, and holds most of the dollars. Current coverage of 1:180 across all segments means nobody is covered well.
Four CSMs at $95k is $380k, cheap against $1.9M, but it still cannot give 62% of logos a meaningful ratio. So allocate by economics:
- $50k-plus and high-potential mid-market: named CSMs at roughly 1:25-40. One saved or expanded account pays for a CSM several times over.
- $15k to $50k: pooled CSM plus tech-touch, with automated onboarding, scaled reviews, and health-triggered outreach.
- Under $15k: digital only, with in-product onboarding milestones, office hours, a community, and automated renewal notices. Do not cut it before checking acquisition cost and self-serve conversion; a packaging change such as a minimum ACV or a lighter-support tier is usually better than exit.
- Fix onboarding first. Seventy-four days to first value against a 30-day target is the cheapest churn reduction available and needs no headcount.
Fund the four CSMs, but assign them by ARR and expansion potential, not logo count.
2. Rebuild the health model. What signals go in, and what governance keeps it honest?
Read the worked answer
Nineteen green accounts churning within 60 days means the score is measuring activity, not risk.
Rebuild it:
- Use separate models per segment. SMB churn is driven by onboarding completion, time to first value, admin login continuity, and payment failures. Enterprise churn is driven by sponsor change, integration depth, renewal timing, and escalations. A blended score hides both.
- Add leading and relational signals: onboarding milestones, seats activated versus purchased, integration count, breadth of use across teams, executive sponsor engagement, support sentiment and escalation history, renewal proximity, invoice status, and champion turnover where observable.
- Weight by evidence, not intuition. Backtest on 24 months of churned and retained accounts and set weights so the model would have flagged at least 70% of churners 90 days out, at a false-positive rate the team can actually work.
- Use bands rather than one number: red, amber, green, each with a required play and a named owner. A score nobody acts on is reporting, not operations.
- Governance: monthly review of missed churners, quarterly re-weighting with documented changes, and a rule that no account is auto-renewed, auto-discounted, or auto-escalated on score alone.
Add a second headline metric beside the score: the percentage of accounts reaching first value within 30 days. It is your earliest reliable predictor.
3. The board asks why NRR dropped and whether 104% is acceptable. What do you say?
Read the worked answer
Give the board the decomposition rather than a defense.
What I would say:
- NRR is 104% because three things moved: gross retention held at 91%, contraction rose, and expansion did not grow fast enough. Put a number on each.
- Mix matters. A fast-growing SMB logo base dilutes NRR mechanically. Twenty-six percent churn on a segment that is 62% of logos but 18% of ARR drags the blend even if enterprise is healthy. Show NRR by segment: enterprise is roughly 124%, combining 95% retention with 131% expansion, while SMB sits well below 100%. The blended figure is a portfolio artifact as much as a performance signal.
- The honest weakness is expansion capacity. With 8% of accounts getting a business review in a year, expansion is luck rather than process.
What I would not do is commit to 115% next year without showing the mechanism: named-account coverage, a library of expansion plays, onboarding time down to 30 days, and the CSM investment. If the board wants the number, price the program that produces it.
Add two disclosures going forward: NRR by segment and a cohort view by signup year. Set targets per segment rather than blended, and report first-value time and review coverage monthly so the board can see next quarter whether the trend is turning.