Your NRR Hides What Your GRR Reveals

A founder I work with runs a 58-person vertical SaaS business at $11M ARR. Her Q1 board deck led with NRR of 108%. The narrative was clean. Expansion is working, customers love us, here's the slide.

A week later her CS lead pulled GRR for the same period. It was 89%. Eleven dollars of every hundred walked out the door in Q1, dressed up by twenty dollars of expansion from the customers who stayed.

Nobody had been hiding anything. The NRR number was real. So was the GRR number. They told completely different stories about the same business.

What the benchmarks now show

Median bootstrapped SaaS companies between $3M and $20M ARR run 103% NRR and 91% GRR. SMB-focused SaaS sits at 97% NRR. Public SaaS median NDR has compressed from 125% in 2021 to 108% in early 2026.

Two things are happening underneath those medians.

Expansion is harder than it was. Seat growth at existing customers slowed when customer hiring slowed. Price increases that ran through annually for five years are getting pushback. The engine that papered over churn is throttling down.

Churn isn't improving. Vendor consolidation, budget cuts, the second look at every renewal. GRR has stayed in the 88-92% range for SMB SaaS through the last three years even as NRR moved around.

What sits between those two numbers is the part of the business that depends on selling more to customers who haven't decided whether to stay.

Why your board will start asking

For most of the last decade, NRR was the trophy metric. Anything over 110% was a company that didn't need to win new logos to grow. Boards loved the number because it priced into multiples cleanly.

That math is changing. When NRR compresses from 130% to 105%, the difference between strong and weak retention is no longer hiding behind expansion. It's sitting on the GRR line. A company at 105% NRR with 95% GRR is a fundamentally different business from one at 105% NRR with 85% GRR. Boards that cared about valuation a year ago are starting to ask the second question.

NRR tells you what your existing customer base did this period. GRR tells you what it would have done if you couldn't sell anyone anything more.

What to actually look at

Pull GRR by cohort and segment for the last six quarters. Most companies haven't done this since the last fundraise. Cohort data shows whether retention is degrading among newer customers, which is the early signal nobody catches in a blended number.

Run the same cut by ACV band. Often the median GRR hides a barbell. High retention in the top twenty accounts, much lower retention in the long tail. Whether that pattern matters depends on where new revenue is coming from.

Then split contraction out from churn. A customer that downgrades from 50 seats to 20 isn't the same problem as a customer that left. They need different fixes. NRR collapses both into one number. So does most board reporting.

For a 30-person SaaS business with 200 customers, this analysis is a half-day of work in the data warehouse and a conversation with CS. Most teams haven't done it because the headline number was good enough to skip past.

It won't be good enough in 2026.