AM Metrics: NRR, GRR, Expansion Contribution, Account Health

Key Numbers for AM Measurement

• NRR formula: (Starting ARR + Expansion - Downgrade - Churn) / Starting ARR
• GRR formula: (Starting ARR - Downgrade - Churn) / Starting ARR — expansion is excluded
• Top-quartile B2B SaaS NRR benchmark: 120%+ for product-led, 110%+ for sales-led (KeyBanc / OpenView surveys)
• Top-quartile GRR benchmark: 90%+ for enterprise, 85%+ for mid-market, 75%+ for SMB
• Forecast accuracy target for a mature AM: renewal forecast within plus or minus 5% of actual for three consecutive quarters
• Lead time target for churn warning: 60+ days between health-score downgrade and customer churn notification

The AM had a book of 32 accounts. Every one of them was green on the dashboard going into Q2. The QBR deck looked clean. By end of quarter, the gross retention number came in at 67%.

Five accounts had churned. Three more had downgraded. None of it was on the health-score timeline. The dashboard said green; the customers signed something else.

The lesson, if you have ever sat in that meeting: green is a story you tell yourself. Renewal is a number the customer signs. The gap between the two is where AM metrics either earn their keep or quietly become decorative.

Why AM Metrics Become Useless

Account management metrics are leading indicators when the rubric is honest, the cadence is weekly, and someone audits health scores against actual outcomes. They become lagging indicators (and useless) when health scores are vanity, when Net Revenue Retention is reported only at the company level, and when no one looks at the data until QBR season.

Every metric on this page is either a forecast or an autopsy. The forecast set tells you what is about to happen so you can change it. The autopsy set tells you what already happened so you can explain it on a board slide. Both are useful. Confusing them is how a team ends up surprised every quarter.

NRR vs GRR: What They Actually Measure

Net Revenue Retention and Gross Revenue Retention answer different questions. Running only one of them hides half the truth.

Net Revenue Retention (NRR) measures whether the existing customer base is growing or shrinking after expansion is counted. It rewards expansion strength.

Gross Revenue Retention (GRR) measures retention quality with expansion stripped out. It exposes whether the base is leaking.

The Formulas

NRR = (Starting ARR + Expansion - Downgrade - Churn) / Starting ARR
GRR = (Starting ARR - Downgrade - Churn) / Starting ARR

GRR is capped at 100%. NRR can exceed 100% when expansion outpaces losses.

Sample Math: A $2M Book

Take a book that started the quarter with $2,000,000 in ARR. During the quarter:

• Expansion ARR: $300,000 (upsell, cross-sell, seat additions)
• Downgrade ARR: $100,000 (seat reductions, plan downgrades)
• Churn ARR: $200,000 (full cancellations)

Run the math:

GRR = ($2,000,000 - $100,000 - $200,000) / $2,000,000
    = $1,700,000 / $2,000,000
    = 85%

NRR = ($2,000,000 + $300,000 - $100,000 - $200,000) / $2,000,000
    = $2,000,000 / $2,000,000
    = 100%

The book is flat on NRR and 85% on GRR. Looked at alone, NRR of 100% sounds adequate. Looked at next to GRR of 85%, the picture changes: the AM is treading water by selling expansion fast enough to cover a leaking base. If expansion slows for one quarter, NRR falls into the 80s and the book contracts.

This is why both numbers are required. NRR alone rewards expansion heroics that mask retention failure. GRR alone hides the engine that is actually working. A healthy book needs GRR above the segment benchmark and NRR above 100% with room to spare.

Segment-Sliced NRR: Where the Average Lies

Company-level NRR averages can hide a broken segment under a strong one. Take a fictional company reporting NRR of 118%:

The 118% headline is real. It is also misleading. SMB is bleeding at 78% and would already be a board-level conversation if anyone reported it on its own. Enterprise expansion is masking the leak. When Enterprise plateaus (and it always does), the SMB problem becomes a company problem with no warning.

Slice NRR by segment, by tenure cohort (under one year, one to three years, three-plus years), and by ICP fit before publishing a single number. The average is not the truth.

For more on driving the expansion side of the equation deliberately, see Account Expansion Mastery.

Designing an Honest Account-Health Rubric

Most account-health scores are vanity. They ask AMs to feel a customer's temperature and pick a color. By the time the color changes from green to yellow, the customer is already drafting a non-renewal email.

An honest rubric scores observable evidence on a defined scale, weights the components by predictive value, and gets audited monthly. Here is a starting structure that holds up in production:

The Four Components

The 1-5 Scoring Scale (with Evidence Prompts)

Each component scores 1 to 5. The score must be defensible with named evidence, not adjectives.

Product usage (40% weight)

• 5: Weekly active user count above 80% of licensed seats; usage trending up over last 90 days
• 4: WAU 60-80% of licensed seats; flat or up
• 3: WAU 40-60% of seats; flat
• 2: WAU 20-40% of seats; flat or declining
• 1: WAU below 20% of seats; declining

Relationship depth (25% weight)

• 5: Three or more named relationships across at least two functions; AM has met economic buyer in last 90 days
• 4: Two named relationships; economic buyer met in last 180 days
• 3: One strong relationship plus one weak; economic buyer name known but not engaged this quarter
• 2: Single point of contact; no economic-buyer relationship
• 1: Single point of contact who has not responded in 30+ days

Commercial signals (20% weight)

• 5: Multi-year contract; auto-renewal active; net 30 payment on time; recent expansion booked
• 4: Annual contract; on-time payment; no procurement escalations
• 3: Annual contract; one late payment or one procurement question this cycle
• 2: Quarterly or month-to-month; payment delays; procurement asking comparison questions
• 1: Late payment unresolved; procurement actively benchmarking competitors; reduction in scope discussed

Executive sponsorship (15% weight)

• 5: Named exec sponsor; attended last QBR; referenced product in internal company communication
• 4: Named exec sponsor; attended one of last two QBRs
• 3: Named exec sponsor; not engaged in last two QBRs
• 2: Sponsor named but no longer in role; no replacement identified
• 1: No exec sponsor identified

The Math

Multiply each score by its weight, sum, and convert to a percentage:

Account: ExampleCo
Usage:        4 x 0.40 = 1.60
Relationship: 3 x 0.25 = 0.75
Commercial:   4 x 0.20 = 0.80
Sponsorship:  3 x 0.15 = 0.45
                 Total = 3.60 / 5.00 = 72%

Color band: Green 80%+, Yellow 60-79%, Red below 60%
ExampleCo = Yellow

A 72% score reads "yellow." Green requires above 80%, which forces at least three components scoring 4 or higher. That is harder to fake than a vibes-based traffic light.

The Calibration Ritual

A rubric without calibration drifts within a quarter. Run this monthly:

1. Pull every account that churned or downgraded in the last 90 days.
2. Pull each one's health score from 90 days before the event.
3. Any score that was green or strong yellow at T-90 is a defect. The AM owns the defect and walks the leadership through what they missed and why.
4. Update the rubric weights or evidence prompts based on the pattern of misses.

Three calibration cycles tighten the rubric to within a few points of actual outcomes. Skip calibration and the rubric quietly returns to vanity within a quarter.

Measuring Expansion Contribution Per AM

Expansion is where comp plans get gamed if the metric is sloppy. Track four numbers, not one:

The Price-Increase Trap

If the comp plan rewards "expansion ARR" without distinguishing source, an AM can hit number by passing through a 7% annual price increase on every renewal. That is renewal pricing power, not expansion. It does not validate product fit, it does not predict next-year retention, and it gives a falsely confident NRR signal.

Split expansion into two buckets in the dashboard:

• Product expansion: new modules, new use cases, additional seats serving new functions
• Price expansion: per-seat price increases, plan-tier increases on the same scope

Compensate product expansion at full rate. Compensate price expansion at a discounted rate or fold it into the renewal target. The AMs working the right motion will out-earn the AMs running price-only books within two cycles.

The Lagging Trio: Renewal Rate, Churn Dollar, Churn Count

Renewal rate, churn dollar, and churn count are autopsies. They tell you what happened. Read them together; never alone.

Read all three together. Example: 12 logos churned representing $400K in ARR. Logo count is high; dollar is moderate. Slice by segment: 11 of 12 are SMB averaging $25K each, one is mid-market at $125K. Verdict: SMB cohort has a fit problem; mid-market has a one-account problem. Different fixes; different urgency.

These three are useful for QBR slides and board updates. They are dangerous as primary AM scorecard metrics because they look backward. Pair them with the leading set below.

The Weekly AM Scorecard

The leading-indicator dashboard belongs in the AM's hands every Monday morning. Here is the layout that works:

WEEKLY AM SCORECARD - Week of [Date]

Quarter Pacing
- NRR pacing: 108% (target 115%, gap = $180K expansion)
- GRR pacing: 91% (target 90%, on pace)
- Expansion pipeline: $480K (3.0x quarterly target)

Book Health
- Green accounts: 18 of 32 (target 20+)
- Yellow accounts: 9
- Red accounts: 5 (representing $620K ARR)

At-Risk
- Renewals at risk this quarter: 2 accounts, $310K
- Health downgrades in last 30 days: 4 accounts
- Accounts with no AM contact in 30+ days: 3 (FLAG)

This Week's Required Actions
- Red account follow-up: 5 conversations scheduled
- Exec sponsor confirmation: 2 outreach emails
- QBR prep: 3 sessions for accounts renewing in 90 days

The discipline is weekly. The scorecard is one screen. If an AM cannot see their pacing in 60 seconds on a Monday morning, the scorecard is too complicated.

For the renewal motion that this scorecard is meant to feed, see Renewal Negotiation: Secure Early. The QBR cadence that drives expansion pipeline lives in QBRs That Drive Expansion.

The Manager 1:1 Metric Prep Sheet

Five numbers, one story per number. That is the structure of every AM-to-manager 1:1 that produces real coaching:

1. NRR pacing this quarter: the one account most likely to move it
2. GRR pacing this quarter: the one account most at risk of churn or downgrade
3. Expansion pipeline coverage: the one opportunity that closes this month
4. Health-score downgrades in the last 14 days: the one account that needs a recovery plan
5. Accounts with no contact in 30+ days: the one that is going to get a call before Friday

If the AM cannot tell the story behind each number in two sentences, the metric is decorative. Manager coaching collapses into reviewing dashboards instead of solving accounts.

Common Pitfalls

A short list, drawn from rebuilds across enough teams to know:

• Vanity health scores: every account green until two weeks before churn
• Reporting NRR only at company level: hiding poor segment performance under whale expansion
• No leading-indicator dashboard: the team only sees metrics in QBR slides
• Counting price-increase ARR as expansion: rewarding the AM for renewal pricing power alone
• Penalizing churn from accounts an AM was assigned 30 days ago: punishing context they did not own
• Health rubrics that score intent: "they say they love us" instead of "logins per week, exec attended last QBR, multi-year contract"

For the broader pattern of what goes wrong in AM execution, see Account Manager Common Pitfalls.

Measuring Success

A mature AM measurement system produces these outcomes:

• Forecast accuracy within plus or minus 5% of actual NRR for three consecutive quarters
• Churn warning lead time: median 60+ days between health-score downgrade and customer churn notification
• Expansion ARR per AM trending up quarter over quarter at the cohort level
• Zero "surprise" churns in the last 12 weeks: every churn was telegraphed by health score and at-risk flag

By week 6 of a 13-week cycle, an AM with a real scorecard should be able to predict their own quarter within 5%. If they cannot, the rubric is wrong, the cadence is too slow, or the metrics are decorative. Fix the system before grading the AM.

How Rework Supports AM Measurement

AMs running real scorecards juggle four data sources: usage telemetry from the product, relationship and exec-sponsor data from CRM, commercial signals from billing, and renewal status from contracts. Most teams stitch these together in a quarterly spreadsheet that nobody opens between QBRs. Rework CRM keeps account health as a live field on every record, with custom rubric weights and the audit trail that calibration cycles need. Pair it with Rework Work Ops to manage the weekly scorecard cadence (at-risk follow-ups, QBR prep tasks, and the "no contact in 30 days" flag) without bouncing between tools. CRM starts at $12 per user per month; Work Ops at $6 per user per month. The job description for the AM role this scorecard is built for: Account Manager.

What to Do This Week

If you are an AM rebuilding your own measurement: pull your last quarter's outcomes, score every account on the 1-5 rubric for the date 90 days before the quarter ended, and find the gap between your old vibes-based score and the rubric's score. That gap is the size of the change you need to make.

If you are a CS leader rebuilding the team's: publish segment-sliced NRR for the last four quarters. Find the cohort that the average is hiding. That cohort is your Q3 priority.

The metrics on this page are not the answer. The discipline of running them weekly is.

Learn More

• Account Expansion Mastery
• QBRs That Drive Expansion
• Account Manager Common Pitfalls
• Renewal Negotiation: Secure Early
• Account Manager Job Description