Prompt and scope
A B2B SaaS product billed monthly or annually sees net revenue retention (NRR) for an existing-customer cohort fall from 108% to 96%. Leadership wants to launch a higher tier immediately, but you must determine whether the problem is churn, downgrades, declining value, or a missing expansion path.
Define NRR on the same opening paid cohort. Separate churn, contraction, expansion, and reactivation, state whether the unit is the account, contract, seat, or usage, and propose product research, experiments, and stop conditions. Do not mix new-customer revenue into NRR. Stripe's MRR growth definition likewise separates new, reactivation, expansion, contraction, churn, and foreign-exchange effects.
What the interviewer evaluates
- Whether you lock the cohort, time window, currency, and billing scope first.
- Whether you turn one percentage change into an actionable revenue bridge.
- Whether you distinguish missing value, blocked expansion, and billing or data errors.
- Whether you connect the metric to product decisions, research, and experiments.
- Whether each lever has leading indicators, guardrails, and stop conditions.
Atlassian's product interview handbook emphasizes customer value, outcomes, communication, and cross-team influence. A strong answer treats NRR as a shared decision surface for product, sales, customer success, and finance.
Clarifying questions before answering
- Is 96% monthly, quarterly, or annual NRR? The window changes noise and intervention time.
- Is the cohort based on first payment, contract start, or renewal date? Are trials and one-time fees excluded?
- Does expansion include added seats, tier upgrades, usage growth, and cross-sell? Is reactivation separate?
- Is the decline concentrated by account size, industry, region, plan, or acquisition channel?
- Could price changes, FX, discounts, refunds, failed billing, or migrations explain it?
30-second answer framework
I would first freeze the cohort and definition, then build a bridge from 108% to 96%: opening revenue minus churn and contraction plus expansion and reactivation. Next I would segment accounts by usage depth, renewal risk, and expansion trigger. If value is not formed, I would repair activation and the core workflow; if value is clear but expansion is blocked, I would test a predictable seat or usage path. Every option gets cohort NRR, retention, usage, and customer-cost guardrails before I promise growth.
Step-by-step deep answer
Step 1: Freeze the metric definition
Let opening cohort revenue be (S) and ending revenue from the same customers be (R). NRR = (R / S). R can only come from the opening cohort; new customers are excluded. Record currency conversion, discounts, taxes, usage settlement, and failed-billing treatment so the number is reproducible.
Step 2: Draw the revenue bridge
Represent ending revenue as (S - churn - contraction + expansion + reactivation + FX). Churn takes a customer to zero revenue; contraction lowers revenue while the customer remains active; expansion adds seats, tier, or usage; reactivation returns a previously zero account. Store each movement by account and month instead of inferring revenue from account counts.
Step 3: Segment to find the break
Calculate NRR, logo retention, ARPA, and usage depth for SMB, enterprise, industry, contract length, and acquisition channel segments. A 96% aggregate may come from a few large downgrades or many early-stage small-account losses. Align cohort curves with key-feature adoption, active seats, successful tasks, and support contacts.
Step 4: Classify the root cause
If customers cannot complete the core workflow, the issue is value formation; improve onboarding, templates, and reliability. If customers use the product deeply but remain on a low tier, the issue is the path from value to expansion; test transparent capacity signals, admin controls, and predictable upgrades. If usage and billing disagree, fix the data and billing pipeline before changing the roadmap.
Step 5: Design research and experiments
For value formation, randomize customers to onboarding or workflow improvements and observe activation, continued use, and 90-day retention. For expansion, test explainable seat or usage prompts and observe expansion MRR, upgrade conversion, and support volume. Stratify by price, contract, and sales touch so sales intervention is not misread as a product effect.
Step 6: Set leading indicators and guardrails
Leading indicators include the share of accounts completing a key job, active-seat ratio, accounts reaching a capacity threshold, and admin invitations. Outcomes include cohort NRR, GRR, logo retention, and expansion MRR. Guardrails include refunds, complaints, support load, performance, margin, and cost from overuse; optimizing only NRR can reward forced upgrades.
Step 7: Handle contract and organization boundaries
Annual contracts can delay product effects until renewal. Sales promises, customer-success plans, and billing must share definitions. Preserve account-level audit events showing whether movement came from product, price, FX, or contract terms. Product owns the decision and validation; finance and customer commitments must remain explicit.
Step 8: Define stop and review conditions
Stop an expansion experiment if NRR rises while core usage falls, refunds rise, or customers are pushed into capacity they do not need. If billing error caused the decline, repair data before comparing trends. If value formation and expansion are healthy but NRR remains low, revisit contract structure, customer mix, or market conditions and rebuild the cohort definition.
Model answer
I would not launch a higher tier just because NRR fell from 108% to 96%. I would freeze the paid cohort, period, currency, and revenue scope; build a bridge for churn, contraction, expansion, reactivation, and FX; then segment by account size and usage depth. If key-job completion is weak, I would repair activation and workflow. If usage is deep but expansion is blocked, I would test explainable seat or usage prompts. I would judge results with cohort NRR, GRR, logo retention, and expansion MRR, while guarding refunds, complaints, support, cost, and margin. Forced upgrades or irreproducible data would stop the work.
Common mistakes
- Add new-customer revenue to NRR → the cohort is contaminated → track only opening customers.
- Read one aggregate NRR → a few large accounts hide the distribution → segment by size, industry, and contract.
- Treat churn and contraction as one event → the fix is misdirected → separate zero revenue from downgrade.
- Push a higher tier when usage is high → usage may only increase cost → check customer outcomes and margin.
- Ignore billing and FX → the roadmap follows false data → retain FX, discount, and billing-failure fields.
- Run only a pricing experiment → value formation stays unknown → measure key jobs and continued use too.
Follow-ups and responses
Does NRR above 100% prove product health?
No. A few expansions can hide broad small-account churn or unsustainable forced usage. Review GRR, logo retention, cohort distribution, and customer outcomes together.
How do you separate expansion from a price increase?
Keep contract version, quantity, unit price, and discount changes. Account separately for price-list, seat, and usage changes, then recompute on one definition.
What if customers use the product heavily but will not upgrade?
Interview buyers and users about measurable outcomes. The issue may be unclear value, budget timing, or plan boundaries; do not assume a more expensive tier is the answer.
When do you repair churn before pursuing expansion?
When GRR or early cohorts keep deteriorating, repair value formation first. Expansion depends on customers choosing to stay and cannot hide foundational churn.
How do you keep sales touches from contaminating a product experiment?
Log sales and customer-success touches, stratify or randomize by touch status, and report product exposure separately from non-product intervention.
Revenue fell while account count stayed flat. What does that suggest?
Check contraction, discounts, lower usage, FX, and failed billing first. Stable logo retention does not guarantee healthy revenue retention.
What result would cancel the expansion roadmap?
If expansion raises revenue while reducing core-job completion, satisfaction, margin, or renewal intent, cancel it and restore customer value and trust.