Retention metric

Gross Revenue Retention. The retention floor.

Gross Revenue Retention is what you actually kept from your existing customers — before any expansion is added back in. It's NRR's stricter cousin. Capped at 100%. And it's the cleanest possible measure of whether your product is genuinely sticky, or whether expansion is masking a churn problem.

What it is

The percentage of starting-period revenue you still have at period end, after churn and contraction — before expansion. Capped at 100%. The honest measure of churn, with no covering up.

Measurement period

12-month rolling.

Compare today's ARR from last year's customers to what those same customers were paying twelve months ago. New customer revenue excluded — GRR is a verdict on the base, not the deals.

Formula

Starting MRR − Churn − Contraction

Starting MRR

× 100

When to review

Monthly.

Weekly is too noisy. Quarterly is too late.

Why it matters

If GRR is bad, nothing else can save you.

GRR is the most honest retention number in your business. NRR can flatter you — one big expansion deal can lift NRR above 100% while customers are quietly walking out the back door. GRR can't be flattered. It only goes down. The only question is how fast.

This is why investors look at GRR right after NRR. If your NRR is 110% and your GRR is 95%, you have a healthy product and a real expansion motion. If your NRR is 110% and your GRR is 78%, you're masking a churn problem with one or two big upsells — and that masking won't hold for long. The gap between the two numbers tells the real story.

At PipelineCRM, our GRR sat at 85–87% for years. We were an SMB product, and SMB customers churn at a rate that's structurally higher than mid-market or enterprise. We weren't going to win on retention. What we tracked instead was whether the number was stable or sliding — because a sliding GRR doesn't mean you've got a sales problem. It means you've got a product or onboarding problem. And those are slower to fix than anything in your pipeline.

GRR is the metric that tells you whether your product is doing its job. If it's trending down, every other fix is downstream of a product or onboarding fix.

Worked example

Three SMB SaaS companies. Same starting line. Different floors.

All three start the year at $1M ARR. None of them add a single new customer. We're measuring only what survives.

Heavy churn

78%

Starting ARR$1,000,000
Churn−$180,000
Contraction−$40,000
Retained$780,000

Heavy SMB churn. Almost a quarter of last year's customers walked. Whatever new logos sales lands this year, a third of them are just backfilling.

Workable

88%

Starting ARR$1,000,000
Churn−$100,000
Contraction−$20,000
Retained$880,000

Typical SMB SaaS. Not a victory, not a crisis. The product is doing its job for most customers, but the bucket is real and you're running on the new-logo treadmill to keep growing.

Sticky product

95%

Starting ARR$1,000,000
Churn−$40,000
Contraction−$10,000
Retained$950,000

Sticky workflow product, often mid-market. Customers don't leave easily. Now every new logo and every expansion dollar compounds on top of a stable base.

Benchmarks

What good looks like, by segment.

Below 80%

Structural problem. Almost always product or onboarding, occasionally a wrong-ICP problem. Sales can't outrun this.

80–90%

Typical for SMB SaaS. Workable but you'll always be running hard to grow. Stable is the priority — sliding is the warning.

90–95%

Healthy mid-market territory. Product is sticky, churn is mostly involuntary (cards, business closures), not preference.

95%+

Enterprise or workflow-essential SaaS. Best-in-class. At this level, GRR isn't a metric you optimize — it's a metric you protect.

When GRR is trending down

Three plays that actually move it.

When GRR slides, the instinct is to send it to customer success and tell them to save more accounts. That's the wrong move. By the time CS gets involved, the customer has already decided. GRR is a product and onboarding metric pretending to be a customer-success metric — so the plays look very different from the NRR playbook.

— 01 Run a cohort test

White-glove ten customers. Self-serve the rest.

Take your next 20 new customers. Manually onboard ten — actual humans, white-glove setup, configuration done for them. Let the other ten self-onboard the way you normally would. Watch both at 30, 60, and 90 days. If white-glove retains meaningfully better, you have an onboarding problem dressed up as a product problem. Most SaaS companies do, and they don't know it because they never run the test.

— 02 Interview the cancellers

Listen for ICP, not features.

Most cancellation interviews go straight to "what feature was missing?" — and most customers give a polite, useless answer. The better question is upstream: were they ever the right customer to begin with? If ten interviews keep surfacing the same pattern — wrong size, wrong industry, wrong workflow — you don't have a churn problem, you have an ICP problem. Most operators don't look here. It's where the biggest fixes live.

— 03 Own activation

The first 90 days are the whole game.

You need a sharp, falsifiable definition of "activated" — a specific event or set of events that means the customer has felt the product working. Not "logged in three times." Something like "imported their data, invited a teammate, completed their first workflow." Then measure what percent hit it by day 14, 30, and 90. Customers who don't activate in the first 90 days almost always churn within 12 months. If activation is under 70% at day 90, fixing that is the highest-leverage move on your scorecard.

Common mistakes operators make with GRR.

Treating GRR drops as a CS problem when they're a product problem.

This is the biggest one. When GRR slides, the first call goes to the head of customer success. But CS has no control over the product, often no control over onboarding, and almost no influence on whether the customer should have bought in the first place. The structural question is who actually owns activation? If the answer is "nobody, or whichever team feels guilty this quarter," your GRR will keep sliding regardless of how many save plays CS runs.

Forcing self-serve onboarding to save CS costs.

The most expensive mistake in SMB SaaS. Companies push customers to configure the product themselves to keep CS headcount low — and then watch GRR erode because half those customers never get to the moment of value. White-glove onboarding feels expensive until you compare its cost to the cost of replacing a churned customer with a new one. Customers who get set up correctly are the ones who stay.

Mixing voluntary and involuntary churn into one number.

A customer cancelling because your product didn't work for them is a very different signal than a customer cancelling because their credit card expired or their business shut down. Track them separately. Involuntary churn is usually a finance fix — better dunning, retry logic, longer grace periods. Voluntary churn is the product/onboarding signal. When you mix them, you lose both signals at once.

Reading one bad month as a trend.

SMB GRR is noisy month-to-month. One month with a couple of big logo losses can swing the number five points. Look at trailing 3-month and trailing 12-month before drawing conclusions. A sliding 12-month trend is a problem. A bad month is usually just a bad month.

Reporting GRR without segmenting by cohort or plan tier.

An 88% blended GRR can hide a 75% GRR on your starter tier and a 96% GRR on your top tier. The blended number tells you nothing about what to fix. Segment by plan, by industry, by company size, by acquisition channel. The segment-level numbers are where the actual decisions live.

How Upbeat helps

Track GRR weekly. Not just on the quarterly board call.

Most SaaS teams only look at GRR when their investor asks. By that point, a problem has been compounding for months. Upbeat pulls GRR (and activation, and cohort retention) from your stack every Monday — and tells you when the trend changes, while you can still do something about it.

See how it works → See pricing

Related metrics

Don't read GRR in isolation.

GRR doesn't fix itself on a quarterly cadence.

Upbeat puts retention on your Monday scorecard — alongside activation, cohort behavior, and the moves to make when one slides.

Email Nick directly →