GTM Operations
New Logos Add. NRR Multiplies. Only One Compounds.
Chase new logos and you are running addition. Net revenue retention runs multiplication, and at 120% versus 102% the base you already own diverges 26x over twenty years. Here is the retention build order, the GRR gap that keeps NRR honest, and a worked example that reconciles to the divergence.
· 13 min read
For years the growth story had one hero: the new-logo chart. Every board deck opened with it, every kickoff celebrated it, every hiring plan was built to feed it. Sign more logos, grow faster. That story broke the quarter someone plotted net revenue retention next to the logo count. New logos are addition. NRR is multiplication, and a valuation multiple pays for multiplication, not addition.
Two companies can post identical new-business numbers and be worth three or four times different amounts, and NRR is most of the reason. The median B2B SaaS company runs about 102% NRR while best-in-class clears 120% (Aleph and Benchmarkit, 2026 SaaS benchmarks). That looks like an 18-point gap on a slide. It is not a gap. It is two different exponents, and over years the two bases pull apart by an order of magnitude. Watch the same starting dollar diverge before you read the argument.
The 102% base stays constant. The field around it grows each level.
That divergence is the whole argument. Hold new logos identical between the two companies and every dollar of the gap is retention. The 102% company is running up a down escalator, selling new business to stand still. The 120% company funds this year’s growth from last year’s book before a single new logo signs. Three things decide which one you become: what NRR measures, why it has to be read next to GRR to stay honest, and the build order that gets both numbers out of the CRM instead of a spreadsheet someone rebuilds by hand once a quarter.
Why the divergence is a magnitude, not a gap
Run two companies side by side. Both start the year at $10M ARR. Both sign the same new logos. One retains at 102% net, one at 120%. The ratio between their bases grows by a factor of 1.176 every year (120 divided by 102), and a factor compounding annually is how you get from an 18-point difference to a 26x difference. The base alone, before either company signs anything new, tells the story.
View as table
| Point | Value |
|---|---|
| Yr 0 | 10M |
| Yr 5 | 24.9M |
| Yr 10 | 61.9M |
| Yr 15 | 154M |
| Yr 20 | 383M |
The zoom at the top and this line chart are the same numbers seen two ways. At year ten the 120% base is $61.9M against $12.2M, a 5.1x spread; by year twenty it is $383M against $14.9M, a 25.8x spread. A valuation multiple is a bet on future cash from the base you already own, which is why the same $10M of revenue is worth several times more at 120% than at 102%. NRR is the number the multiple prices, and it prices it exponentially.
The benchmarks worth taping to the wall (Aleph and Benchmarkit, 2026): median B2B SaaS lands near 102% NRR, the top quartile clears 110%, and best-in-class runs 120% and up. On the gross side, median GRR sits near 84%. That distance between median NRR and median GRR is the entire expansion story of the category, and it is exactly where most of the risk hides.
GRR and NRR answer different questions
The two metrics look similar and hide different truths, so conflating them is how a leaky business tells itself a happy story.
Gross revenue retention measures how much recurring revenue you keep before any expansion. Churn and downgrades only. It can never exceed 100%.
GRR = (starting ARR − churn − downgrades) / starting ARR
GRR is the honest floor. It tells you how leaky the bucket is when nobody is topping it off.
Net revenue retention adds expansion back in.
NRR = (starting ARR − churn − downgrades + expansion) / starting ARR
NRR can and should clear 100%. It tells you whether growth from the base outruns the leaks. You need both, because NRR alone lies by omission. A company can post 110% NRR while churning 20% of its base if a few big accounts expand enough to paper over the loss. That is concentration risk wearing a good number, and GRR strips the makeup off.
Here is how to read the two numbers together against the benchmarks. Print it and tape it next to the divergence chart.
| Read | GRR | NRR | What it means |
|---|---|---|---|
| Elite | 92% | 122% | Sticky base, strong expansion, funds its own growth |
| Healthy | 90% | 112% | Above both floors, the profile investors pay up for |
| Category median | 84% | 102% | Barely multiplying, expansion just covers the leak |
| Papered-over | 78% | 108% | NRR looks fine, a leaky bucket propped by a few whales |
| Bleeding | 80% | 92% | Running up a down escalator, new sales refill the leak |
The two rows that should worry you are the papered-over and the bleeding lines. Same NRR band as healthy companies in one case, but the GRR underneath tells you the retention is borrowed, not earned. On the divergence math above, the difference between the healthy row and the category-median row is the difference between the top line and the bottom line of that chart.
A worked example that reconciles to the divergence
Start the year at $10M ARR. Over twelve months three accounts cancel outright for $800K, existing customers cut spend by $400K, and existing customers add $1.6M through upsell and seat growth.
GRR = ($10M − $800K − $400K) / $10M = $8.8M / $10M = 88%
NRR = ($10M − $800K − $400K + $1.6M) / $10M = $10.4M / $10M = 104%
Read them together. NRR at 104% looks fine on a slide, and it sits just above the 102% category median (Aleph and Benchmarkit, 2026). GRR at 88% is above the 84% median but under the 90% floor a healthy SaaS business wants, which means the bucket leaks faster than it should and the 104% is expansion carrying churn on its back.
Now split that $1.6M of expansion by account, because the blended number hides where it comes from.
| Source of expansion | Accounts | Expansion ARR | Share |
|---|---|---|---|
| Top five accounts | 5 | $1,150K | 72% |
| Long tail | 40+ | $450K | 28% |
| Total expansion | $1,600K | 100% |
Seventy-two percent of all expansion sits on five logos. Pull those five and expansion falls to $450K, so NRR drops to ($10M − $800K − $400K + $450K) / $10M = $9.25M / $10M = 92.5%. The 104% is not the base’s health. It is five renewals you cannot afford to lose. Map that back to the divergence chart: 104% blended puts you near the category-median line that reaches $14.9M in twenty years, but the true base health without the whales is 92.5%, which decays below the starting line and never compounds at all. The concentration is the difference between the flat bottom line and a line that bends downward.
Put your own numbers in and watch the two levers move against each other. Churn and expansion sit the same distance from 100%, but they are not the same problem, and the divergence chart weighs them differently over time.
net retention
Above 100% the existing book grows on its own before a single new logo signs. Below 100% you are running up a down escalator: new sales have to refill the leak before they add anything.
net retention: 110.0%
Where the four components come from
The reason NRR usually lives in a hand-built spreadsheet is that most CRMs store net change, not the components. You can back into NRR from net change, but you can never compute GRR, which means you threw away the more honest of the two numbers. The fix is to store four things separately, per account, per period.
The instrument is a report, not a hand-built spreadsheet. Once the four fields exist per account per period, both metrics fall out of one query. This is the artifact I run every quarter; it never computes net change, so GRR survives.
-- Both metrics from four stored components, cohorted by start quarter.
-- Never store net change: you can recover NRR from it but never GRR.
SELECT r.start_quarter,
SUM(r.starting_arr) AS start_arr,
SUM(r.churn_arr) AS churn,
SUM(r.downgrade_arr) AS downgrade,
SUM(r.expansion_arr) AS expansion,
-- GRR uses the first three components only, caps at 100%
(SUM(r.starting_arr) - SUM(r.churn_arr) - SUM(r.downgrade_arr))
/ NULLIF(SUM(r.starting_arr), 0) AS grr,
-- NRR adds expansion back in, can and should clear 100%
(SUM(r.starting_arr) - SUM(r.churn_arr) - SUM(r.downgrade_arr) + SUM(r.expansion_arr))
/ NULLIF(SUM(r.starting_arr), 0) AS nrr
FROM retention_components r
WHERE r.period_end >= DATEADD(month, -12, CURRENT_DATE)
GROUP BY r.start_quarter
ORDER BY r.start_quarter;Two clauses carry the honesty. GROUP BY r.start_quarter is the cohorting from step 2, so no vintage hides inside a blended average. The separate churn_arr and downgrade_arr columns are why GRR exists at all: strip them into a single net field and the query can still return NRR while GRR becomes uncomputable.
The retention build order
This is the order I stand up the instrument so both numbers come straight out of the CRM, cohorted, every quarter, with no hand math. Each step earns its place before the next one lights up, the same way each year of the divergence chart depends on the one before it.
- 1
1. Store four fields, never net change
Add Starting_ARR, Churn_ARR, Downgrade_ARR, and Expansion_ARR as period-stamped fields, or a small child object, one row per account per quarter. In the worked example those hold $10M, $0.8M, $0.4M, and $1.6M. Store only the net $0.4M and you can compute NRR but never GRR, because you cannot recover churn and downgrade from a net figure.
- 2
2. Cohort by start date
Tag every account with the quarter it entered the book. Measure a cohort at month zero and again twelve months later. Blend every vintage into one company-wide ratio and you hide which cohorts retain and which quietly rot; a 2025 cohort at 96% GRR and a 2026 cohort at 80% average to a comfortable-looking 88% that describes neither.
- 3
3. Save both formulas as report metrics
GRR = (start − churn − downgrade) / start = $8.8M / $10M = 88%. NRR = (start − churn − downgrade + expansion) / start = $10.4M / $10M = 104%. Define both as saved formulas on the report type so nobody recomputes them by hand and nobody quietly changes the denominator.
- 4
4. Report the pair side by side
Every retention slide shows GRR and NRR next to each other, per cohort. The gap here is 16 points, 104 minus 88, and that gap is your expansion story. The wider it runs, the more of your growth rides on a handful of expanding accounts rather than a sticky base.
- 5
5. Add the concentration check
Compute what share of expansion comes from your top five accounts. In the worked example that is $1.15M of $1.6M, or 72%. Pull those five and NRR falls from 104% to 92.5%. When NRR clears 110% on 70%-plus top-five concentration, the good number is borrowed; flag it before the board reads NRR as safety.
Build the steps in that order and the divergence chart stops being a hypothetical. You can see, per cohort, which vintages sit on the compounding line and which sit on the flat one, and you can name the five accounts whose renewal decides which line the whole book rides.
Blended versus cohorted, and why it matters
The single company-wide NRR number is the one most teams report and the one that hides the most.
| One blended NRR | Cohorted GRR + NRR pair | |
|---|---|---|
| What you see | One number: 104% | GRR 88% and NRR 104% per vintage |
| Churn visibility | Hidden inside the net figure | Isolated, trends on its own line |
| Concentration risk | Invisible until a whale churns | Top-5 expansion share flagged |
| Which cohorts rot | Averaged away | Named by start quarter |
| Twenty-year base | Assumed to compound | Traced to the line it truly rides |
Where it lands
New logos add. NRR multiplies, and over twenty years multiplication turns an 18-point benchmark gap into a 26x difference in the base you own. Track GRR and NRR together, because NRR tells you whether you are growing and GRR tells you whether that growth stands on a sticky base or on one renewal you cannot afford to lose. Instrument the four components separately so you can compute both honestly, cohort by start date, and read the gap between them as your risk gauge.
Start with one schema change: add the four fields, backfill last year from the renewal records you already have, and put GRR next to NRR on the next board slide. If the two numbers sit more than 20 points apart, you have a retention conversation to lead before the quarter leads it for you. For the pipeline-side version of this same discipline, see why 3x coverage lies to you, and for the expansion motion that turns a sticky base into a compounding one, see why expansion cannot be left to chance. Get NRR above 110% with GRR above 90% and you own an asset that funds itself. Everything else in GTM is feeding it.
Keep reading
One email. Every week.
One email a week: an operating problem I solved or botched, with the model, the numbers, and what I would change. No roundups, no theory, unsubscribe whenever it stops being useful.
The newsletter opens soon.
Connect a provider in src/config.ts