Skip to content
YibudYibud

Before you trust an NRR slide

SaaS GRR (Gross Revenue Retention): the floor under NRR before you trust expansion

A board slide that prints “NRR is 118%” without a GRR column is asking you to trust expansion you have not yet named. Gross revenue retention — also called gross dollar retention / GDR — asks a smaller, honester question: of the recurring revenue that was already there, how much survived without counting upsells? Write the floor before you look. Hitting or missing it is a decision signal. It is not a forecast that you will succeed.

Last updated: November 2, 2026

Direct answer

What is SaaS GRR — and why write the tripwire first?

GRR (gross revenue retention / gross dollar retention) = (starting recurring revenue from a named cohort − churn − contraction, with expansion excluded) ÷ starting recurring revenue, over a named window — commonly 12 months. It cannot exceed 100%. It answers how much of the base survived without counting upsells, not whether you will succeed. Write the tripwire before you look: “for cohort or segment S as of date D, trailing-12-month GRR must stay ≥ X% — or we stop scaling acquisition, cut the segment, or fix contraction before we trust NRR.” Hitting or missing it is a decision signal, not a prediction of success.

Key takeaways

What to remember

  • NRR includes expansion. GRR excludes it. NRR is always ≥ GRR. GRR and logo retention cannot exceed 100%; NRR can. ChartMogul’s fictional SaaS Inc.: start MRR 770, end MRR from the same cohort 800 → NRR 800/770 ≈ 103.9%. Exclude expansion and the same cohort ends at 630 → GRR 630/770 ≈ 81.8%. New logos and reactivations outside the starting cohort do not enter the retention calc.
  • High NRR can hide a leaky GRR. ChartMogul: NRR of 105% could be 90% GRR + 15% expansion, or 70% GRR + 35% expansion. Without GRR you cannot see gross churn and contraction. Keith Wallington’s board framing, as ChartMogul quotes it: put GRR, NRR, and growth on one page so the room sees new business, retention, and upsell as one journey — not a single net percentage.
  • Bessemer, Scaling to $100 Million (21 September 2021): gross retention is the share of revenue maintained after turned-off or downgraded revenue, and it does not account for any expansion. By ARR range it sits relatively consistently at 85–90%. Their lesson framing for strong retention is about 85% gross / about 120% net. Successful SMB cases can sit at 70–80% — HubSpot in its 3 years pre-IPO, in their note — then watch CLTV / CAC. NRR depth lives on the NRR page; this page keeps only the contrast (about 105–145% at $1–10MM ARR → about 105–125% at $100MM+; averages about 140% → 120%).
  • Bessemer, State of the Cloud 2019 (6 February 2019), segment aims by ACV: SMB under $12K → gross 70–80%, net 80–100%. Mid-market $12–50K → gross 80–90%, net 90–120%. Enterprise $50K+ → gross >90%, net >100%. Churn is a symptom. If you sit outside the band, investigate the leaky bucket — do not paste HubSpot’s 70–80% onto an enterprise book and call it a law.
  • Write the tripwire before you measure: as-of date + segment + starting recurring base + window (prefer 12 months) + statistic + threshold + named action if missed (stop paid acquisition for that segment / pause expansion storytelling until GRR recovers / stop selling the segment). Always report GRR next to NRR. A Yibud score is not GRR and not a success prediction. Pre-product, do not “optimize GRR” by starving discovery.

Why it matters

Why NRR can look fine while the base is leaking

Founders treat a triple-digit NRR as permission to scale. Expansion is a real motion. It is also a blanket. If you never name the dollars that left or shrank, you cannot tell whether the existing book is holding — or whether a few upsells are covering a hole you already paid CAC to fill.

NRR can mask churn

ChartMogul’s 2023 retention report is blunt: an NRR of 105% could be 90% GRR plus 15% expansion, or 70% GRR plus 35% expansion. Same net slide. Different companies. Without GRR you have no clue how high or low gross churn and contraction really are. That is why this page exists even when NRR “looks fine.”

Logo retention is not the same sitting

ChartMogul: if revenue concentrates in a few accounts, logo retention and GRR can split hard. Imagine 100 customers a year ago and 20 leave — logo retention is 80%. If those 20 held half the revenue, GRR is 50%. Ten cheap logos staying does not save a whale cancel. Count logos on the logo-churn page. Weigh dollars here.

The floor is a spend gate, not a trophy

Bessemer’s Scaling piece treats high gross revenue retention as hard to do without if you want a successful cloud business. That is a portfolio observation, not your personal odds. Write the floor first. If you miss it, stop telling the expansion story until the base holds — or stop scaling that segment. Matching 85–90% does not prove product-market fit.

The five terms

GRR, versus NRR, versus logo retention, the 100% ceiling, and the cohort window

These five fields are the instrument. If any one is fuzzy, the number is a slide, not a tripwire. This page does not publish a universal “good GRR is 90%” grade. Use Bessemer’s ARR-band and 2019 segment bands, and ChartMogul’s quartile / ARPA sittings, then write your own floor.

GRR — starting recurring revenue kept, expansion excluded

Gross Revenue Retention / GRR (also gross dollar retention / GDR) is the share of a named starting recurring base still there after churn and contraction, with expansion left out. Bessemer: it tells you what percentage of revenue you maintained; it nets out customers who turned off or downgraded; it does not account for any expansion. ChartMogul: if you start at $100k MRR, what share of that same-cohort revenue remains 12 months later, excluding expansion.

Versus NRR — expansion in, and NRR is always at least GRR

NRR / net dollar retention adds expansion back in. NRR is always ≥ GRR. NRR can print above 100% when upsell outruns churn plus contraction. GRR cannot. The NRR page owns First Round’s Level 2–4 NRR gates and the full Bessemer NRR table. This page keeps NRR only as the neighbor that can hide a leak. Report both. Never board-slide NRR alone.

Versus logo retention — accounts versus dollars

Logo / customer retention counts accounts that stayed. GRR weights by recurring revenue. ChartMogul’s 80% logo / 50% GRR split is what happens when the logos that left held half the book. Do not call an 80% logo-retention number GRR. Do not copy Gabe’s ICP / SMB Good–Better–Best logo-churn tables onto this page — those live next door.

The ceiling — GRR cannot exceed 100%

Because expansion is out, the honest ratio cannot print above 100%. If your “gross” number wants to, you included expansion — that is NRR, not GRR. This page skips ChartMogul’s growth-band sentence that labels a GRR “over 100%” sitting; by definition GRR ≤ 100%. We keep their coherent figures: GRR >85% growing 1.5–2.5× faster; 60–75% band averaging 20.2% / year; <60% band 13.4% / year.

Cohort window — named start base, named period, usually 12 months

Name the customers (or dollars) live at period start. Name the window. ChartMogul and Bessemer both treat a 12-month sitting as the common clock — it works for monthly and annual contracts. New business and reactivations outside that starting cohort stay out. A 90-day sketch, annualized, is a slogan. Write the window before the first report.

The arithmetic

The formula — and ChartMogul’s SaaS Inc. worked example

There is no Yibud GRR formula. Use the starting-base sitting the sources actually wrote. The worked example below is ChartMogul’s fictional SaaS Inc., not a Yibud invention and not a benchmark you should copy as a target.

Gross revenue retention

GRR = (starting recurring revenue − churn − contraction) ÷ starting recurring revenue. Expansion excluded. Cap at 100%.

Starting recurring revenue is the book of customers live at period start. Churn left. Contraction stayed and paid less. Do not add expansion. Do not fold this month’s new logos into the starting base. If the ratio wants to print above 100%, you are no longer computing GRR.

ChartMogul’s fictional SaaS Inc. — their numbers, their label

Sid Jain / ChartMogul, SaaS Retention Report 2023 (1 November 2023). They label this SaaS Inc. as a worked example, not a live company. One year ago: four paying customers, total MRR 770. Today, from that same group: some churned, some contracted, some expanded. Total MRR today from the same group is 800. New business and reactivation outside the starting cohort do not enter.

NRR on the same cohort — 800 / 770 ≈ 103.9%

ChartMogul: end MRR from the same customers who existed a year ago is 800. NRR = 800 / 770 ≈ 103.9%. That net number includes expansion. It can print above 100%. It does not tell you what left.

GRR on the same cohort — 630 / 770 ≈ 81.8%

ChartMogul: excluding expansion, MRR today from that same group is 630. GRR = 630 / 770 ≈ 81.8%. Same start. Same customers. A different, honester floor. The 103.9% NRR would have hidden that 81.8% if you only printed the net.

What this example is not

It is not Bessemer’s 85–90% ARR-band sitting. It is not HubSpot’s 70–80% SMB vignette. It is not a Yibud “good GRR” grade. Copy the method: same starting cohort, expansion out of GRR, new logos out of both. Do not paste 81.8% onto your book as a target.

If you only remember one crossing: 103.9% NRR and 81.8% GRR can be the same fictional company in the same year. Always put both on the dated page.

Bessemer, 2021

ARR-band gross retention — and the HubSpot SMB exception

Mary D’Onofrio, Ethan Ding, and Atlas editors, Scaling to $100 Million (21 September 2021). Gross retention on a dollar basis is the share of revenue maintained; it nets out turned-off or downgraded revenue; it does not account for any expansion. The rows below are what that page named. They are not a Yibud letter grade.

Bessemer Scaling to $100 Million — gross retention sittings, 21 September 2021
SittingWhat they namedHow to label it
By ARR rangeGross retention relatively consistent at 85–90%.Portfolio sitting, not a seed law, not proof of PMF.
Why it still mattersHard to build a successful cloud business without high gross revenue retention.Their judgment on the metric’s weight — not your personal odds.
SMB exception — HubSpotSuccessful SMB cloud businesses can sit at 70–80% gross retention. Their example: HubSpot in its 3 years pre-IPO. Then keep a close eye on CLTV / CAC.Segment exception, source-tagged. Not a universal SMB law, and not an enterprise target.
Strong-retention lesson framingAbout 85% gross / about 120% net.Lesson 1 framing in Scaling. Net ranges stay on the NRR page.

NRR contrast only, as a neighbor pointer: Bessemer’s net retention ranges about 105–145% at $1–10MM ARR, decreasing to about 105–125% at $100MM+; averages about 140% → 120%. This page does not rebuild that table or First Round’s L2–L4 NRR gates. Read them on /saas-nrr. CLTV / CAC depth lives on the LTV:CAC page.

Bessemer, 2019

Segment gross (and net) aims by ACV

Byron Deeter, State of the Cloud 2019 (6 February 2019). Retention “has different meanings depending on customer segments.” The bands below are aims they wrote for a cloud company, not a 2026 market census and not a Yibud grade table. If you sit outside the band, they say investigate the leaky bucket — churn is a symptom.

Bessemer State of the Cloud 2019 — gross and net retention aims by ACV segment
SegmentAverage contract valueGross retention aimNet retention aim
SMBUnder $12K70–80%80–100%
Mid-market$12–50K80–90%90–120%
Enterprise$50K+>90%>100%

2019 aims, labeled as aims. Do not flatten them with Scaling’s 85–90% ARR-band sitting into one homemade “good GRR.” Do not treat HubSpot’s 70–80% SMB exception as permission for an enterprise book to live at 75%.

ChartMogul, 2023

Quartiles, ARPA, and growth — compare like with like

Sid Jain, SaaS Retention Report 2023 (1 November 2023). Anonymized aggregates from 2,100+ SaaS businesses, over 2022, limited to companies active the full 12 months. These are ChartMogul’s dataset, not a law for your fourteen-person company, and not a Stripe or KeyBanc table.

Top-quartile GRR by ARR — more harmonized than NRR

Because GRR excludes expansion, top-quartile values sit closer across ARR bands than NRR does. ChartMogul: $3–8m ARR → top-quartile GRR 81.2%. $15–30m ARR → 83.8%. The bottom quartile improves as companies scale — they attribute that to more businesses reaching product-market fit. That is their reading of their sample, not a promise that your GRR will rise if you wait.

Best-in-class at any stage — over 86%

ChartMogul: best-in-class gross retention at any stage is over 86%. You can only afford to lose 14% of your gross revenue in a year — their wording, their ceiling, not a Yibud guarantee. Clearing 86% is a sitting against their bar. It does not prove you will succeed.

ARPA splits — do not compare apples to oranges

Top quartile with ARPA over $500 / month hit 90%+ GRR. Top quartile with ARPA under $50 / month only hit 60–70%. Only 5.3% of businesses with ARPA under $10 / month have GRR over 85%. 35.7% of businesses with ARPA over $500 / month have GRR over 85%. Best-in-class GRR is about 95% for mid-market / enterprise sellers; B2C tops out around 75–80%. Judge your number against a motion that looks like yours.

Growth link — GRR over 85%, not a homemade median

ChartMogul (they exclude under $3m ARR from that growth analysis): companies with GRR over 85% grow at least 1.5–2.5× faster. The 60–75% GRR band averaged 20.2% growth per year. The under-60% band averaged 13.4% per year. Association in their 2022 sample, not a forecast that your GRR will manufacture growth.

Board page — GRR + NRR + growth, not NRR alone

Keith Wallington, investor and chairperson, as ChartMogul quotes him: if you chart GRR, NRR, and growth rate on a single page, you have what you need for a board meeting — it gets people thinking about the whole customer journey: new business, retaining it, upselling it. Paraphrased here, attributed there. A lonely NRR percentage is not that page.

This page omits ChartMogul’s growth sentence that bands GRR “over 100%.” GRR cannot exceed 100% by definition. We do not invent a fix. We keep the coherent >85% / 1.5–2.5× / 20.2% / 13.4% figures.

How to run it

How to run a GRR tripwire without faking scale

This is Yibud’s suggested sequence, not a Bessemer checklist and not a ChartMogul mandate. Honor the tripwire you wrote before the first cohort report. If you cannot name the decision the number would change, you are building a slide.

  1. 1

    Write the tripwire before you look

    As-of date. Segment. Starting recurring base. Window — prefer 12 months. Statistic (trailing GRR %). Threshold you will honor — Bessemer’s 85–90% ARR-band, the 2019 segment band that matches your ACV, ChartMogul’s >86% best-in-class line, or a tighter floor you justify in writing. Named action if missed: stop paid acquisition for that segment, pause expansion storytelling until GRR recovers, or stop selling the segment. Date the page. Do not rewrite the bar after the close.

  2. 2

    Lock the starting base and keep expansion out

    Customers live at period start. Their recurring revenue that day is the denominator. Churn and contraction leave the numerator. Expansion stays out of GRR. New logos and reactivations booked during the window stay out of both GRR and NRR. If you only have 90 days, report a 90-day GRR and say so — do not multiply by four.

  3. 3

    Compute GRR first, then NRR on the same base

    Subtract churned and contracted recurring revenue from the start base, divide, cap at 100%. Then compute NRR on the same start, same window, with expansion in. ChartMogul’s 105% = 90+15 or 70+35 warning is the reason. Put both numbers on the dated page. Never board-slide NRR alone.

  4. 4

    Split by segment and ARPA before you average

    Bessemer 2019 treats SMB, mid-market, and enterprise as different aims. ChartMogul treats ARPA under $50 / month and over $500 / month as different worlds. A blended company GRR can hide a cheap-logo leak or a whale cancel. Show n, start ARR, and the segment. Do not paste HubSpot’s 70–80% onto an enterprise book. Do not paste 85–90% onto a $9 ARPA consumer motion and call it failure.

  5. 5

    Honor the action. If you are pre-product, use a different instrument.

    If the window closes and you missed the bar you wrote, do the thing you named. Do not invent next quarter’s expansion so NRR covers the miss. No paying cohort yet? Do not “optimize GRR.” Use interviews, a willingness-to-pay test, a smoke test, or kill criteria first. A Yibud score is not GRR.

What you leave with

The GRR tripwire (copy this)

If the sitting produced a percentage and no dated action, you did not finish. Fill the brackets. Leave nothing as “retention is fine” or “NRR will cover it.”

The fill-in contract

Product: [name]. Cohort / segment S: [ICP / SMB / mid-market / enterprise / ARPA band], written before the first report. As-of date D: [ ]. Starting recurring base: [ARR or MRR $ ] for customers live that day. Window: [prefer 12 months], books close [date]. Hypothesis this number can falsify: [the existing book holds without counting upsells, so we can scale acquisition / trust the NRR story]. Definition: dollar GRR (expansion excluded; not logo retention). Churned $: [ ]. Contraction $: [ ]. Expansion $: [ ] — reported, excluded from GRR. GRR: [ ]. NRR on the same base: [ ]. Threshold: [e.g. trailing-12-month GRR ≥ 85% (Bessemer ARR-band / lesson framing), or the 2019 segment band that matches our ACV, or >86% if we wrote ChartMogul best-in-class as the gate — pick one and justify; not HubSpot’s 70–80% pasted onto enterprise, not a homemade 90% for all SaaS]. Action if we miss: [stop paid acquisition for this segment / pause expansion storytelling until GRR recovers / stop selling the segment / fix contraction first]. Action if we clear: [keep going / next vintage on the same definitions]. A Yibud score does not fire this line.

The “≥85% / 2019 band / >86%” menu is Yibud’s suggested vocabulary from Bessemer and ChartMogul — not a law, and not permission to skip NRR. Rewrite the threshold before you measure if you have a reason. Do not rewrite it after.

Example tripwire (Yibud’s suggested template, not a law)

If, for segment S as of 1 January, trailing-12-month dollar GRR is below 85% when we close that vintage on 31 December → we do not 3× paid acquisition in Q1; we pause the expansion story until GRR recovers and we name the weakest churn or contraction assumption. Report NRR in the next column. Do not annualize a Q1 window. Do not swap logo retention in after the close. Do not include expansion in the “gross” calc.

A 88% GRR on a 12-month window is still one vintage. A Yibud report score does not fire this tripwire. If you have no paying customers yet, this template is the wrong first instrument.

The fork

When to use a GRR tripwire — and when not to

GRR is the wrong first instrument when there is no recurring revenue yet. It is the right instrument when a paying cohort already exists and you are about to trust NRR, scale a channel, or tell yourself the base is fine because a few accounts expanded.

Use it when you have a paying cohort and an NRR-or-scale decision

You can name starting recurring revenue for customers live at period start, and you can count churn, contraction, and expansion on that same base. You are about to 3× a channel, print NRR without GRR, or treat HubSpot’s 70–80% as a license. That is when a dated GRR floor earns its keep. Prefer a 12-month window when you have it. Always sit GRR next to NRR.

Do not start here when you are pre-product or still need discovery

No paying cohort? There is no honest starting base. Do not invent a forecast GRR. Still testing whether anyone has the problem, will pay, or would be very disappointed? Use interviews, a willingness-to-pay test, a smoke test, the Ellis PMF survey, or kill / pivot / critical-assumption contracts first. Once you have a paying cohort with recurring revenue, write the GRR tripwire for that cohort or segment. Do not starve discovery to decorate an 85% slide. A Yibud score is not GRR.

Not the same page

GRR vs NRR, logo churn, retention curves, Quick Ratio, and the neighbors

Several Yibud pages sit next to this one. Mixing them up turns expansion, account counts, usage flatten, a same-period flow ratio, or a waitlist into “GRR is 88%, so scale.”

NRR — expansion included; this page is the floor underneath

That page owns net revenue retention, First Round’s L2–L4 NRR gates, and Bessemer’s full NRR table. This page owns the floor that NRR can hide. A 118% NRR with an 82% GRR is not the same company as a 118% NRR with a 96% GRR. Read NRR there. Compute GRR here. Do not copy that NRR table onto this page.

SaaS NRR →

Expansion mix — this period’s new bookings, not the retained base

That page asks what share of net new ARR came from existing customers versus new logos. GRR asks how much of the starting base survived without counting those expansion dollars. A 40% expansion mix does not tell you GRR is 88%. An 88% GRR does not tell you this quarter’s bookings were 40% expansion.

SaaS expansion ARR →

Logo churn — account count, not dollar retention

That page asks how many paying accounts left. This page asks how much starting recurring revenue survived without expansion. ChartMogul’s 80% logo / 50% GRR split is why both sittings exist. This page does not copy Gabe’s ICP / SMB Good–Better–Best tables.

Logo churn for startups →

Retention curve — usage over days, not a dollar floor

That page plots whether a sign-up cohort still does the core action at day N. GRR is recurring-revenue kept over a named window, usually 12 months. A flatten at 35% weekly logs does not tell you GRR is 88%. An 88% GRR does not tell you the usage curve flattened.

Retention curves for startups →

Quick Ratio — same-period inflow ÷ outflow, not the retained base

That page asks whether this period’s new + expansion outran churned + contraction (a multiple). GRR is the share of the starting base that remained, expansion excluded. A 4× Quick Ratio can still sit on a leaky GRR. Pair them. Do not average them.

SaaS Quick Ratio →

Vanity metrics — hygiene class, not this floor

That page is the sorting hat: hits, cumulative sign-ups, and waitlist size feel like progress. A GRR with no named starting ARR, no “expansion excluded,” and no tripwire is vanity in a nicer font. “Starting ARR” and “exclude expansion” have to be defined or the number fails the test.

Vanity metrics for startups →

Time to Value — the onboarding clock, not later dollar retention

That page asks how long until a named first outcome. GRR asks whether the dollars from a paying cohort were still there a year later, without counting upsells. Slow TTV can feed later churn. They are still different instruments.

Time to Value for startups →

LTV:CAC and CAC payback — unit economics that a leaky GRR wrecks

Those pages ask whether a cohort’s contribution cleared a multiple of CAC, or how many months recover CAC. Bessemer’s HubSpot note sends you there when SMB GRR sits at 70–80%. A 3.2 LTV:CAC does not tell you GRR. An 88% GRR does not tell you paid CAC pays back.

Magic Number, burn multiple, Rule of 40, Efficiency Score — growth and cash, not this floor

Those pages ask how efficiently S&M turned into incremental revenue, how much cash you burned per net-new ARR, whether growth plus profit cleared 40, or whether ARR YoY plus FCF margin cleared Bessemer’s score. GRR is the retained base. Do not borrow their numbers onto this page.

Discovery and decision contracts — earlier instruments

Willingness to pay, the Ellis PMF survey, kill criteria, pivot criteria, and critical-assumption validation write money, disappointment, Stop, rewrite-one-dimension, or the one claim that would kill the idea. GRR is later, and only after a paying cohort exists. Use those pages first when you are still pre-product.

Where Yibud fits

Neither GRR nor a Yibud score predicts success

Yibud is a free, no-signup startup idea validator. Scores come from a deterministic rule engine, not from a language model guessing success. Optional AI text, when it is used, only polishes prose. It does not invent the number. A Startup MRI report can name a weak monetization or retention assumption, then a Day-7 Continue / Refine / Re-test / Stop signal. The score is a flashlight on the weakest assumption to test next — not GRR, not NRR, not a forecast. GRR work sits next to that assumption: write the floor, do not hope a 118% NRR slide means the bucket does not leak. If you already have a report, start with the named assumption, not the overall score. Then run the cohort yourself. The method on this page stands alone if you never open the analyzer.

Analyze my idea →

Common mistakes

What usually wastes a GRR week

Reporting NRR without GRR

ChartMogul: 105% NRR could be 90% + 15% or 70% + 35%. a lonely net percentage cannot tell you which. Put GRR on the dated page. Keith Wallington’s board framing wants GRR, NRR, and growth together. Do not let expansion cover a leak you already paid CAC to fill.

Treating HubSpot’s 70–80% SMB sitting as a law for enterprise

Bessemer’s Scaling note is a segment exception — HubSpot in its 3 years pre-IPO — then watch CLTV / CAC. Bessemer 2019 aims enterprise ACV $50K+ at >90% gross. Copying 70–80% onto a $50K+ book is how you lower the bar after the leak.

Copying 85–90% as proof of product-market fit

Bessemer’s ARR-band sitting is relatively consistent at 85–90%. ChartMogul notes bottom-quartile GRR improves as more of their sample reaches PMF — their reading, not a certificate. Matching 85–90% means that vintage cleared a bar you wrote. It does not prove PMF. It does not predict success.

Mixing logo retention with GRR

ChartMogul’s 80% logo / 50% GRR split is the warning. Ten $2K logos staying can print a kinder logo number than one $80K cancel. Label logos as logos. Weigh dollars as GRR. Do not copy logo-churn Good / Better / Best tables here.

Including expansion in the “gross” calc

If expansion is in, you are computing NRR. GRR cannot exceed 100%. A “GRR” of 103% is a definition error, not a trophy. ChartMogul’s SaaS Inc. only reaches 103.9% on the net sitting. The gross sitting is 81.8%.

Inventing an industry median — or a universal 90%

This page does not publish a universal “good GRR is 90% for all SaaS.” It does not import Stripe’s later medians, KeyBanc’s NRR tables, or a homemade average of Bessemer and ChartMogul. Use the source-tagged bands. Write your own floor. Do not flatten 2019 ACV aims, 2021 ARR-band consistency, and 2022 ChartMogul quartiles into one letter grade.

Sources

Where these ideas come from

  • Mary D’Onofrio, Ethan Ding & Atlas editors / Bessemer Venture Partners, “Scaling to $100 Million,” 21 September 2021 — Used for the gross-retention definition (share of revenue maintained; nets out turned-off or downgraded revenue; does not account for any expansion); 85–90% by ARR range; hard to build a successful cloud business without high GRR; HubSpot 3 years pre-IPO at 70–80% then watch CLTV / CAC; lesson framing about 85% gross / about 120% net; NRR contrast only (about 105–145% at $1–10MM ARR → about 105–125% at $100MM+; averages about 140% → 120%). This page does not copy the full NRR table.
  • Byron Deeter / Bessemer Venture Partners, “State of the Cloud 2019,” 6 February 2019 — Used for segment aims by ACV: SMB under $12K → gross 70–80%, net 80–100%; mid-market $12–50K → gross 80–90%, net 90–120%; enterprise $50K+ → gross >90%, net >100%; churn as a symptom; investigate the leaky bucket if you sit outside the bands.
  • Sid Jain / ChartMogul, “SaaS Retention Report 2023,” 1 November 2023 — Used for the GRR definition; fictional SaaS Inc. 770 / 800 / 630 (NRR ≈ 103.9%, GRR ≈ 81.8%); 105% NRR = 90+15 or 70+35; 80% logo vs 50% GRR when churned logos held half the revenue; Keith Wallington board framing; 2,100+ SaaS, 2022 aggregates, full-12-month companies; top-quartile GRR 81.2% at $3–8m ARR and 83.8% at $15–30m; best-in-class >86% / 14% loss; ARPA 90%+ vs 60–70%; 5.3% / 35.7% shares above 85%; ~95% vs ~75–80% best-in-class by motion; GRR >85% grow 1.5–2.5× faster; 60–75% band 20.2% / year; <60% band 13.4% / year. Omits the inconsistent “GRR over 100%” growth sentence.

In one paragraph

Summary you can quote

GRR (gross revenue retention / gross dollar retention) is the share of starting recurring revenue kept after churn and contraction, excluding expansion, over a named window — commonly 12 months. It cannot exceed 100%. It answers how much of the base survived without counting upsells, not whether you will succeed. Bessemer Scaling (21 September 2021): 85–90% by ARR; HubSpot 3 years pre-IPO 70–80% then watch CLTV / CAC; strong-retention framing about 85% gross / about 120% net. Bessemer 2019: SMB / mid-market / enterprise aims 70–80% / 80–90% / >90% gross by ACV. ChartMogul 2023 (2,100+ SaaS, 2022): SaaS Inc. 770 / 800 / 630; NRR 105% can be 90+15 or 70+35; top-quartile 81.2% / 83.8%; best-in-class >86%; ARPA splits 90%+ vs 60–70%; GRR >85% grow 1.5–2.5× faster. Write as-of date + segment + start base + window + threshold + action before you look. Always report GRR next to NRR. Matching the floor is a decision signal. Neither GRR nor a Yibud score predicts success.

FAQ

Questions founders actually ask

What is SaaS GRR (gross revenue retention)?

The share of starting recurring revenue you keep after churn and contraction, excluding expansion, over a named window — commonly 12 months. Also called gross dollar retention / GDR. It cannot exceed 100%. ChartMogul’s fictional SaaS Inc.: 630 / 770 ≈ 81.8% GRR versus 800 / 770 ≈ 103.9% NRR. It is a retention-floor tripwire, not a success score. Write the bar before you measure.

What is a good GRR for a startup?

This page does not publish a universal “good GRR is 90%” table. Bessemer Scaling (2021): relatively consistent 85–90% by ARR; SMB exception 70–80% (HubSpot 3 years pre-IPO) then watch CLTV / CAC. Bessemer 2019: SMB 70–80%, mid-market 80–90%, enterprise >90% gross, by ACV. ChartMogul 2023: top quartile 81.2% at $3–8m ARR and 83.8% at $15–30m; best-in-class >86%; ARPA over $500 / month top quartile 90%+, under $50 / month only 60–70%. Write your own tripwire for your segment. Matching a band is not PMF and not a forecast.

How is GRR different from NRR and from logo churn?

NRR includes expansion and can exceed 100%. GRR excludes expansion and cannot. NRR is always ≥ GRR. Logo / customer retention counts accounts. ChartMogul: 80% logo retention can sit next to 50% GRR if the logos that left held half the revenue. Report GRR next to NRR. Label logos as logos.

Can GRR exceed 100%?

No. Expansion is out of the calc, so the honest ratio cannot print above 100%. If it wants to, you included expansion — that is NRR. This page skips ChartMogul’s growth sentence that labels a GRR “over 100%” band, because it conflicts with that definition. We keep their >85% / 1.5–2.5× / 20.2% / 13.4% figures.

Is a Yibud score the same as GRR? Does hitting 85% predict success?

No and no. Yibud scores come from a deterministic rule engine on your idea inputs. They point at the weakest assumption to test next. They are not GRR, not NRR, and not a forecast. Neither an 85–90% Bessemer band, a ChartMogul >86% best-in-class line, nor a Yibud report score predicts success. You do not need an account to use this page.

Write the GRR floor before you trust the NRR slide

Yibud scores weak assumptions with a deterministic engine. You write as-of date, segment, start base, window, threshold, and action while you are still honest — then a year of GRR is a tripwire, not a 118% NRR slide with no floor underneath.

Generate a Yibud report →