MRR Churn: How to Calculate It, Diagnose It, and Fix It

Most SaaS founders I work with at $3M to $15M Annu­al Recur­ring Rev­enue can quote their MRR. Far few­er can tell me what their MRR churn was last month, cal­cu­lat­ed the way an acquir­er would cal­cu­late it. The ones who can usu­al­ly quote the wrong num­ber — they hand me a cus­tomer churn rate dressed up in dol­lar signs, or a gross fig­ure that hides an expan­sion engine, or a month­ly fig­ure they mul­ti­plied by 12 and called annu­al. The MRR churn num­ber sit­ting on the aver­age $5M ARR board deck is wrong by enough to change the con­clu­sion the board reach­es from it.

That mat­ters because mrr churn is the sin­gle most impor­tant reten­tion sig­nal in a SaaS busi­ness. New logo acqui­si­tion can paper over a leaky base for a few quar­ters. It can­not paper over it for­ev­er. Soon­er or lat­er — usu­al­ly around the $8M to $10M ARR mark, when the absolute dol­lars walk­ing out the door start to rival the absolute dol­lars walk­ing in — the leak shows up in the growth rate and the board starts ask­ing why. The point of mea­sur­ing MRR churn clean­ly is to see that gap years before it becomes a growth prob­lem.

This guide walks through what MRR churn actu­al­ly mea­sures, the exact for­mu­las for gross and net, a worked exam­ple with real­is­tic num­bers at the $5M ARR scale, the five mis­takes founders make that dis­tort their num­ber, the bench­marks acquir­ers actu­al­ly use, and a 90-day play­book to move your num­ber in the right direc­tion. By the end you will know what to put on a board slide, what to ignore on a board slide, and which lever to pull when the num­ber is wrong.

What MRR Churn Actually Measures

MRR churn mea­sures the per­cent­age of your exist­ing recur­ring rev­enue base that you lose or fail to retain dur­ing a defined peri­od — usu­al­ly a month. That def­i­n­i­tion has three parts that founders skip over and pay for lat­er.

First, MRR churn is a rev­enue met­ric, not a logo met­ric. It is denom­i­nat­ed in dol­lars of Month­ly Recur­ring Rev­enue (MRR), not in head­count of cus­tomers. One enter­prise cus­tomer at $40,000 a month who down­grades to $10,000 a month is invis­i­ble to your cus­tomer churn rate — they did not can­cel — but they show up imme­di­ate­ly and painful­ly in MRR churn. Twen­ty small cus­tomers can­celling at $200 a month each costs you $4,000; one big cus­tomer can­celling at $4,000 costs you the same. MRR churn treats them equal­ly because the dol­lars are equal. Cus­tomer churn does not.

Sec­ond, MRR churn is mea­sured against an exist­ing base, fixed at the start of the peri­od. The cohort is the cus­tomers you had on day one of the month. Any­one you sign on day two or lat­er does not count toward Start­ing MRR, and their rev­enue does not count toward expan­sion or any­where else in the for­mu­la. The ques­tion the met­ric answers is nar­row on pur­pose: what is hap­pen­ing to the rev­enue I already had? New logo rev­enue is a dif­fer­ent ques­tion, answered by a dif­fer­ent met­ric.

Third, MRR churn comes in two fla­vors that are not inter­change­able: gross and net. The gross ver­sion counts only the dol­lars walk­ing out the door (can­cel­la­tions and down­grades). The net ver­sion sub­tracts expan­sion and reac­ti­va­tion from that gross num­ber. The gross fig­ure tells you how leaky the buck­et is. The net fig­ure tells you whether the buck­et is still fill­ing up despite the leak. You need both, on the same page, every month. Buy­ers will ask for both. Show­ing one with­out the oth­er is the SaaS equiv­a­lent of show­ing rev­enue with­out mar­gin.

If you take only one thing from this sec­tion, take this: a healthy SaaS com­pa­ny at $5M ARR is usu­al­ly run­ning about 1% to 2% month­ly gross MRR churn and zero to slight­ly neg­a­tive month­ly net MRR churn. Any­thing mate­ri­al­ly worse than that, at this stage, is a prob­lem worth inter­rupt­ing your roadmap to fix.

The MRR Churn Formula

The MRR churn for­mu­la has four com­po­nents, plus a denom­i­na­tor. Get the com­po­nents right and the arith­metic falls out clean­ly.

Gross MRR Churn = (Churned MRR + Con­trac­tion MRR) ÷ Start­ing MRR

Net MRR Churn = (Churned MRR + Con­trac­tion MRR − Expan­sion MRR − Reac­ti­va­tion MRR) ÷ Start­ing MRR

Let me define each piece in plain Eng­lish before we plug in num­bers.

• Start­ing MRR — the recur­ring rev­enue from the cohort of cus­tomers you had on day one of the mea­sure­ment peri­od. Not new cus­tomers acquired dur­ing the peri­od. Fixed at the start.
• Churned MRR — recur­ring rev­enue lost from cus­tomers in the cohort who ful­ly can­celled dur­ing the peri­od. Their MRR goes to zero.
• Con­trac­tion MRR — recur­ring rev­enue lost from cus­tomers in the cohort who stayed but reduced their spend (down­grad­ed plan, cut seats, dropped a mod­ule).
• Expan­sion MRR — addi­tion­al recur­ring rev­enue from cus­tomers in the cohort (upsells, cross-sells, seat addi­tions, price increas­es applied to exist­ing cus­tomers).
• Reac­ti­va­tion MRR — recur­ring rev­enue from pre­vi­ous­ly churned cus­tomers who came back dur­ing the peri­od. Most oper­a­tors include this in expan­sion; I pre­fer break­ing it out because reac­ti­va­tion behav­ior tells you some­thing dif­fer­ent from an upsell, and you want to see both sig­nals clean­ly.

The cohort is fixed at the start of the peri­od. A cus­tomer you sign on day two is a new logo, not an expan­sion. Their rev­enue belongs in your new MRR line, not in any churn cal­cu­la­tion. This rule is the sin­gle most vio­lat­ed rule in SaaS report­ing, and it is the rule that lets a fast-grow­ing com­pa­ny qui­et­ly hide reten­tion rot inside aggre­gate num­bers for sev­er­al quar­ters before any­one notices.

A few mechan­ics worth flag­ging before you reach for a cal­cu­la­tor. Use month­ly recur­ring rev­enue, not Annu­al Con­tract Val­ue (ACV) or trail­ing rev­enue, when cal­cu­lat­ing month­ly MRR churn. Annu­al num­bers are fine for year-over-year com­par­i­son, but the math gets slop­py when con­tracts span par­tial peri­ods. And do not annu­al­ize a month­ly MRR churn rate by mul­ti­ply­ing by 12 — the cor­rect annu­al­iza­tion uses geo­met­ric com­pound­ing, which I will show below. Mul­ti­ply­ing by 12 over­states annu­al churn by rough­ly 5 to 25 per­cent­age points depend­ing on the month­ly rate. That is enough to change every con­clu­sion you draw from the num­ber.

A Worked Example at $5M ARR

Gener­ic for­mu­las blur togeth­er. Num­bers do not. Walk through this slow­ly, because the same arith­metic will run on your busi­ness the moment you close this tab.

You run a ver­ti­cal SaaS busi­ness that end­ed April 2026 at exact­ly $416,667 in MRR (that is $5,000,000 ARR ÷ 12). For sim­plic­i­ty assume the entire MRR comes from a sta­ble cohort of cus­tomers who were all active on May 1. May 1 is day one of your mea­sure­ment peri­od. Through May, the fol­low­ing move­ments hap­pen on that cohort — and only on that cohort, ignor­ing May new logos:

Movement	Dollar Amount	% of Starting MRR
Starting MRR (May 1 cohort)	$416,667	—
Churned MRR (full cancellations)	$8,333	2.0%
Contraction MRR (downgrades, seat cuts)	$2,083	0.5%
Expansion MRR (upsells, seat adds)	$12,500	3.0%
Reactivation MRR (returned churned customers)	$1,250	0.3%

Plug into the gross MRR churn for­mu­la:

Gross MRR Churn = ($8,333 + $2,083) ÷ $416,667 Gross MRR Churn = $10,416 ÷ $416,667 Gross MRR Churn = 2.5%

That is the leak­i­ness of the buck­et. Two and a half per­cent of your exist­ing rev­enue base walked out the door in May, between full can­cel­la­tions and down­grades. Now the net:

Net MRR Churn = ($8,333 + $2,083 − $12,500 − $1,250) ÷ $416,667 Net MRR Churn = (−$3,334) ÷ $416,667 Net MRR Churn = −0.8%

Neg­a­tive. That is the right sign. Even though 2.5% of your base leaked out, expan­sion and reac­ti­va­tion more than refilled it. The cohort is, on net, pay­ing you 0.8% more in May than in April — with zero con­tri­bu­tion from new logo sales. That is a self-fund­ing base, which is exact­ly what an acquir­er wants to see.

Trans­late the net num­ber into the met­ric every investor will ask you about: net rev­enue reten­tion. The two num­bers are arith­metic mir­rors:

Net Rev­enue Reten­tion (NRR) = 100% − Net MRR Churn

In our exam­ple: 100% − (−0.8%) = 100.8% month­ly NRR.

Now annu­al­ize prop­er­ly, with com­pound­ing (not mul­ti­pli­ca­tion):

Annu­al NRR = (1 + Month­ly NRR Growth)^12

(1.008)^12 = 1.1003. Annu­al NRR ≈ 110.0%, or equiv­a­lent­ly, an annu­al net MRR churn of approx­i­mate­ly −10.0%. That is a num­ber that opens doors in a fundraise.

To make the com­par­i­son sting, hold every oth­er com­po­nent con­stant and swap the expan­sion and churn lines so churn dou­bles and expan­sion halves:

Movement	Healthy Cohort	Leaky Cohort
Starting MRR	$416,667	$416,667
Churned MRR	$8,333 (2.0%)	$16,667 (4.0%)
Contraction MRR	$2,083 (0.5%)	$4,167 (1.0%)
Expansion MRR	$12,500 (3.0%)	$6,250 (1.5%)
Reactivation MRR	$1,250 (0.3%)	$625 (0.15%)
Gross MRR Churn	2.5%	5.0%
Net MRR Churn	−0.8%	3.4%
Monthly NRR	100.8%	96.6%
Annual NRR (compounded)	110.0%	65.6%

Look at the annu­al NRR col­umn. The healthy cohort, left alone, grows 10% a year before you sell a sin­gle new logo. The leaky cohort, left alone, shrinks by 34.4% a year. To net to zero growth, the leaky cohort has to replace one-third of its entire rev­enue base in new logos every year just to stand still. That is the kind of tread­mill that ends fundrais­ing rounds.

The sin­gle biggest lever in SaaS eco­nom­ics is not new logo acqui­si­tion. It is the math hid­den in this table.

Gross MRR Churn vs Net MRR Churn — When to Lead With Which

These are not inter­change­able. They answer dif­fer­ent ques­tions, and pre­sent­ing one with­out the oth­er is how founders acci­den­tal­ly mis­lead their own boards.

Dimension	Gross MRR Churn	Net MRR Churn
What it counts	Cancellations and downgrades only	Cancellations and downgrades minus expansion and reactivation
What question it answers	How leaky is the bucket?	Is the bucket still filling up despite the leak?
Healthy range at $5M ARR	1% to 2% monthly	−1% to +1% monthly
What it hides	Nothing — it is the gross number	Severity of the underlying leak
What it reveals	Product, pricing, and onboarding problems	Expansion engine health and overall cohort economics
When to lead with it	Internal retention reviews, product roadmap decisions	Board meetings, investor updates, M&A diligence

Lead with gross MRR churn when the audi­ence is inter­nal and the ques­tion is “what is bro­ken.” Prod­uct fit issues, bad onboard­ing, mis­priced con­tracts, a fea­ture gap that drove a key seg­ment to a com­peti­tor — all of these show up in gross first. A net num­ber can be flat or neg­a­tive even when gross is high, because expan­sion is mask­ing it. If you only look at net, you will miss the struc­tur­al prob­lem until your expan­sion engine slows down for an unre­lat­ed rea­son and the gross loss­es sud­den­ly become vis­i­ble.

Lead with net MRR churn when the audi­ence is exter­nal and the ques­tion is “is the cohort self-fund­ing.” Investors, acquir­ers, and lenders care about whether your exist­ing cus­tomer base, on its own, pro­duces more rev­enue next month than it did this month. That is the ques­tion net answers. A net MRR churn of zero or below is the clean­est sig­nal a SaaS busi­ness can send. It says the cus­tomers you already have will fund their own growth, and any new logo dol­lars you add on top com­pound on a base that is itself grow­ing.

The mis­take to avoid is pre­sent­ing only one num­ber. A board deck with only net MRR churn looks great until some­one asks “what is your gross?” and the answer reveals an expan­sion engine paper­ing over a 6% gross leak. A board deck with only gross looks alarm­ing until some­one cal­cu­lates net and sees the cohort is actu­al­ly self-fund­ing. Always show both. Always show them on the same page. Always show the trail­ing six months so the tra­jec­to­ry is vis­i­ble.

The Five Mistakes That Distort Your MRR Churn Number

Across two decades of coach­ing SaaS founders, these are the five errors I see most often. Each one of them, in iso­la­tion, can change your report­ed MRR churn by a full per­cent­age point or more — enough to flip your num­ber from healthy to alarm­ing, or worse, from alarm­ing to false­ly healthy.

1. Mul­ti­ply­ing month­ly MRR churn by 12 to get an annu­al num­ber. Month­ly and annu­al churn are not lin­ear mul­ti­ples. The cor­rect con­ver­sion uses geo­met­ric com­pound­ing: Annu­al Churn = 1 − (1 − Month­ly Churn)^12. At 2% month­ly gross churn, naive mul­ti­pli­ca­tion gives 24% annu­al; the cor­rect answer is 21.5%. At 5% month­ly the gap is 60% vs 46%. At 10% it is 120% vs 71.8%. The com­pound­ing error always over­states annu­al churn, which means founders who use the wrong for­mu­la scare them­selves need­less­ly — until they present the num­ber to a buy­er who cor­rects them, at which point they look like they do not under­stand their own met­rics. Use the com­pound for­mu­la every time.
2. Count­ing new MRR in the wrong buck­et. A cus­tomer signed on May 5 is a new logo, peri­od. Their May rev­enue belongs in your new MRR line. It does not belong in expan­sion, it does not off­set churn, and it does not change Start­ing MRR. This rule sounds obvi­ous. It is vio­lat­ed con­stant­ly, espe­cial­ly by founders who run a sin­gle dash­board that lumps “all MRR added in May” into one num­ber. The fix is to build the four-com­po­nent break­down — start­ing, churned, con­trac­tion, expan­sion, reac­ti­va­tion, new — and report each line sep­a­rate­ly. If your billing sys­tem can­not pro­duce that break­down, treat that as a P1 finance prob­lem to fix this quar­ter.
3. Using ACV-based math for month­ly cal­cu­la­tions. Annu­al Con­tract Val­ue is a per-deal annu­al­ized fig­ure. MRR is a month­ly run-rate fig­ure. They are com­put­ed from the same con­tracts but answer dif­fer­ent ques­tions, and mix­ing them inside a month­ly churn cal­cu­la­tion pro­duces garbage. If you have a $60,000 ACV con­tract that start­ed March 15, the May con­tri­bu­tion is $5,000 of MRR (1/12 of ACV), not $60,000 and not zero. Build the month­ly run-rate clean­ly from the con­tract terms — most billing plat­forms expose an MRR view already, and if yours does not, it is worth a one-week engi­neer­ing project to pro­duce one.
4. Treat­ing con­trac­tion as churn, or vice ver­sa. A cus­tomer who down­grades from $5,000 a month to $3,000 a month is not a can­cel­la­tion. They con­tributed $2,000 of con­trac­tion MRR and $3,000 of retained MRR. Lump­ing that $2,000 into Churned MRR (instead of Con­trac­tion MRR) over­states your can­cel­la­tion rate and under­states your down­grade rate, which makes it hard­er to diag­nose which prob­lem is actu­al­ly worse. Pric­ing-dri­ven down­grades and prod­uct-fit-dri­ven can­cel­la­tions are dif­fer­ent fail­ure modes with dif­fer­ent fix­es; keep them in dif­fer­ent buck­ets so you can see them.
5. For­get­ting to break out reac­ti­va­tions. Reac­ti­va­tion MRR — cus­tomers who churned in a pri­or peri­od and came back — is gen­uine­ly good news, but it is dif­fer­ent from organ­ic expan­sion. A high reac­ti­va­tion num­ber can mean your prod­uct is sticky enough that peo­ple come back after they try alter­na­tives; it can also mean your sales team is win­ning back cus­tomers who churned for fix­able rea­sons. Both are use­ful sig­nals, both are dif­fer­ent from “exist­ing cus­tomers buy­ing more,” and both are worth a sep­a­rate line in the report. Most oper­a­tors bury reac­ti­va­tion inside expan­sion. I break it out by default.

There is a sixth mis­take worth a brief men­tion: defin­ing the cohort incon­sis­tent­ly from one month to the next. If May’s Start­ing MRR is the cus­tomers active on May 1, June’s Start­ing MRR has to be the cus­tomers active on June 1 — which is May 1’s cohort minus May churn plus May new logos, where May new logos now count as “start­ing” for June. Doc­u­ment the def­i­n­i­tion once and apply it every month. The worst ver­sion of any met­ric is the one that qui­et­ly rede­fines itself between quar­ters.

Benchmarks: What MRR Churn Should Look Like at Your Stage

The right MRR churn num­ber depends on where you are. Ask­ing “what is a good MRR churn rate” with­out a stage qual­i­fi­er pro­duces use­less answers, because the same num­ber is excel­lent at one scale and alarm­ing at anoth­er. The table below shows what I see across the SaaS com­pa­nies I coach and what buy­ers typ­i­cal­ly expect at each stage.

ARR Stage	Healthy Monthly Gross MRR Churn	Healthy Monthly Net MRR Churn	Annual NRR (Compounded)
Pre-$1M ARR (early)	3% to 5%	1% to 3%	70% to 90%
$1M to $3M ARR	2% to 3%	0% to 2%	80% to 100%
$3M to $10M ARR	1.5% to 2.5%	−0.5% to +1%	95% to 115%
$10M to $25M ARR	1% to 1.5%	−1% to 0%	100% to 115%
$25M+ ARR	< 1%	< −0.5%	110% to 130%+

A few impor­tant notes on this table. First, the num­bers reflect month­ly rates, not annu­al. Mul­ti­ply cor­rect­ly when con­vert­ing. Sec­ond, the ranges are delib­er­ate­ly wide because end mar­kets vary — a low-ACV SMB SaaS busi­ness should expect high­er gross churn than a high-ACV enter­prise SaaS busi­ness at the same ARR stage, sim­ply because small­er cus­tomers churn more often. Third, the “healthy net MRR churn” col­umn assumes a func­tion­al expan­sion motion. If you have no expan­sion motion at all (no upsell path­way, no seat expan­sion, no price lad­der), your net will track your gross, and these bench­marks will look unat­tain­able.

The most impor­tant num­ber on this table is the net col­umn at the $10M to $25M ARR stage. Once you cross $10M ARR, acquir­ers and lenders start pric­ing your busi­ness on the assump­tion that the exist­ing cus­tomer base is, on net, self-sus­tain­ing. If your net MRR churn is pos­i­tive at that stage, your ter­mi­nal val­ue gets dis­count­ed heav­i­ly. If it is at or below zero, you get full cred­it for the dura­bil­i­ty of the cohort, which is typ­i­cal­ly the largest sin­gle dri­ver of val­u­a­tion mul­ti­ple at that scale.

(A note on time-sen­si­tive data: the bench­mark ranges above reflect typ­i­cal 2026 SaaS mar­ket con­di­tions. Spe­cif­ic num­bers vary by ver­ti­cal, end mar­ket, and com­pet­i­tive dynam­ics. The ranges are includ­ed to show rel­a­tive dif­fer­ences between stages rather than as absolute tar­gets — your peer set may run hot­ter or cool­er. Always cross-check against ver­ti­cal-spe­cif­ic bench­marks before set­ting a board-lev­el tar­get.)

What Drives MRR Churn — and What Does Not

There is a pop­u­lar nar­ra­tive in SaaS that cus­tomer suc­cess teams are the lever for reten­tion. They are not the lever. They are the indi­ca­tor that the lever has been pulled.

The actu­al dri­vers of MRR churn, in rough order of impact, are these:

• Pric­ing fit. Cus­tomers churn when the price exceeds the per­ceived val­ue at their stage. A cus­tomer signed up for a $2,000/month plan because it was the cheap­est option, and then real­ized they were only using 30% of the fea­ture set, will down­grade or can­cel. The fix is not a cus­tomer suc­cess call; the fix is a bet­ter-shaped pric­ing lad­der.
• Onboard­ing com­plete­ness. Cus­tomers who reach “first val­ue” (what­ev­er the action is that deliv­ers the core ben­e­fit) in the first 14 to 30 days churn at a frac­tion of the rate of cus­tomers who do not. The sin­gle high­est-lever­age reten­tion invest­ment most ear­ly-stage com­pa­nies can make is in onboard­ing instru­men­ta­tion: mea­sure how many cus­tomers hit first val­ue, set an inter­nal tar­get, and treat any drop below the tar­get as a P1 issue.
• Prod­uct-stage fit. A cus­tomer at $2M ARR has dif­fer­ent needs than a cus­tomer at $20M ARR. If your prod­uct is built for the $2M crowd and you let your sales team sell into the $20M crowd, you will see those large cus­tomers churn with­in 6 to 12 months because the prod­uct can­not scale with them. The fix is sales-side dis­ci­pline, not cus­tomer suc­cess hero­ics.
• Account sta­bil­i­ty. A cus­tomer whose inter­nal cham­pi­on leaves the com­pa­ny with­in 90 days of the deal clos­ing is at mate­ri­al­ly high­er churn risk. There is no soft­ware fix for this; the fix is rela­tion­ship-den­si­ty — mak­ing sure more than one per­son on the cus­tomer side is bought in before the deal clos­es.
• Cus­tomer suc­cess motion. Impor­tant, but down­stream of the four items above. A great cus­tomer suc­cess team can move reten­tion by maybe 2 to 3 per­cent­age points on the mar­gin. A bad pric­ing lad­der will cost you 10.

The impli­ca­tion is uncom­fort­able for many founders. If your MRR churn is too high, the answer is rarely “hire anoth­er CSM.” The answer is usu­al­ly a struc­tur­al change to pric­ing, sales tar­get­ing, or onboard­ing — work that is hard­er, slow­er, and more polit­i­cal than adding head­count. The CSM addi­tion is what you do after you have fixed the upstream caus­es, to com­pound the gains.

A 90-Day Playbook to Bring MRR Churn Down

If you fin­ish read­ing this and your gross MRR churn is mean­ing­ful­ly above the bench­mark for your stage, here is the sequence I would run if I were sit­ting in your CEO chair. This is the same sequence I work through with coach­ing clients when their num­ber is wrong.

Days 1–14: Diag­nose. Pull the last 12 months of MRR move­ments and break each month clean­ly into the five com­po­nents — start­ing, churned, con­trac­tion, expan­sion, reac­ti­va­tion. Then seg­ment each com­po­nent by cus­tomer cohort (ACV band or ver­ti­cal) and by tenure (under 90 days, 90 days to 12 months, 12+ months). The out­put of this two-week sprint is a sin­gle sheet that shows where the loss­es are con­cen­trat­ed: which seg­ment, which tenure band, which loss type. Until you can see that sheet, every “reten­tion ini­tia­tive” is guess­work.

Days 15–45: Pick one dri­ver and fix it. Based on the diag­nos­tic, pick the sin­gle largest con­trib­u­tor — pric­ing fit, onboard­ing com­ple­tion, prod­uct-stage fit, account sta­bil­i­ty, or CSM motion. Pick one. Try­ing to fix three at once is how 90-day ini­tia­tives become 90-week ini­tia­tives. Build a spe­cif­ic inter­ven­tion: a pric­ing lad­der restruc­ture, an onboard­ing com­ple­tion track­er, a sales play­book change, an exec­u­tive spon­sor pro­gram, a CSM staffing mod­el. What­ev­er it is, scope it small enough to ship in 30 days and mea­sur­able enough to eval­u­ate in 60.

Days 46–75: Mea­sure the inter­ven­tion. Track the spe­cif­ic cohort that expe­ri­enced the inter­ven­tion and com­pare its churn behav­ior to the pri­or trail­ing 90 days. Do not aggre­gate this with the rest of the base — the effect will get washed out. The sig­nal you are look­ing for is a clear drop in the fail­ure mode you tar­get­ed, in the cohort you tar­get­ed. If the sig­nal is not there, the inter­ven­tion did not work; cut it and try a dif­fer­ent one. If the sig­nal is there, pre­pare to scale it.

Days 76–90: Scale or restart. Either roll the work­ing inter­ven­tion across the broad­er base (with the same mea­sure­ment dis­ci­pline so you can see con­tin­ued impact) or, if noth­ing worked, return to the diag­nos­tic and pick a dif­fer­ent dri­ver. Either way, the goal of day 90 is to have a piece of evi­dence on a sin­gle sheet of paper that says “this inter­ven­tion moved MRR churn from X to Y in this cohort.” That sheet of paper is what turns reten­tion from a vague aspi­ra­tion into a man­aged lever.

Three guardrails through­out. First, do not change the met­ric def­i­n­i­tion while you are work­ing on the num­ber — it is the eas­i­est way to fool your­self into think­ing the inter­ven­tion worked. Sec­ond, do not stack inter­ven­tions on top of each oth­er; you will not know which one moved the num­ber. Third, do not stop mea­sur­ing. The com­pa­nies that bring churn down sus­tain­ably are the ones that treat reten­tion as a per­ma­nent met­ric, not a one-quar­ter project.

Frequently Asked Questions

What is the dif­fer­ence between MRR churn and cus­tomer churn?

MRR churn is denom­i­nat­ed in dol­lars of month­ly recur­ring rev­enue; cus­tomer churn is denom­i­nat­ed in head­count of cus­tomers. They can diverge sharply. A SaaS busi­ness with high con­cen­tra­tion in a few large accounts can have low cus­tomer churn (most small cus­tomers stick around) and high MRR churn (one big cus­tomer left). Always look at both. For pric­ing and prod­uct deci­sions, MRR churn is the more impor­tant sig­nal. For top-of-fun­nel and brand deci­sions, cus­tomer churn mat­ters more. The full break­down is in the SaaS cus­tomer suc­cess met­ric guide.

Is gross MRR churn or net MRR churn the more impor­tant met­ric?

Gross is the lead­ing indi­ca­tor; net is the lag­ging indi­ca­tor. Gross tells you when some­thing breaks in the prod­uct, pric­ing, or onboard­ing — it moves first. Net tells you whether the cohort as a whole is still self-fund­ing — it moves lat­er, and it is what acquir­ers price. Run both. Inter­nal reten­tion dis­cus­sions should usu­al­ly start with gross; exter­nal investor and buy­er dis­cus­sions should usu­al­ly lead with net. Show­ing one with­out the oth­er is the most com­mon mis­take in SaaS report­ing.

What is a good month­ly MRR churn rate?

It depends on your ARR stage. Pre-$1M ARR, 3% to 5% month­ly gross MRR churn is nor­mal — you are still fig­ur­ing out prod­uct fit. By $5M ARR, healthy is 1.5% to 2.5%. By $10M+ ARR, healthy is under 1.5% gross and at or below zero on net. Any­thing mate­ri­al­ly worse than those ranges, at your stage, is worth treat­ing as a struc­tur­al prob­lem rather than a sea­son­al blip.

How do I con­vert month­ly MRR churn to annu­al MRR churn?

Use the com­pound for­mu­la, not mul­ti­pli­ca­tion: Annu­al Churn = 1 − (1 − Month­ly Churn)^12. Mul­ti­ply­ing month­ly by 12 over­states the annu­al rate by 5 to 25 per­cent­age points depend­ing on the lev­el. At 2% month­ly the cor­rect annu­al is 21.5%, not 24%. At 5% month­ly the cor­rect annu­al is 46%, not 60%. The com­pound­ing effect mat­ters for any deci­sion that depends on the absolute lev­el of the annu­al fig­ure — fundrais­ing mate­ri­als, M&A mod­els, inter­nal fore­casts.

How is MRR churn dif­fer­ent from net rev­enue reten­tion?

They are arith­metic mir­rors. NRR = 100% − Net MRR Churn. If your net MRR churn is −1.0%, your NRR is 101.0%. They con­tain the same infor­ma­tion and are inter­change­able in any rig­or­ous dis­cus­sion. The con­ven­tion is that oper­a­tors talk about MRR churn (they want the num­ber to go down) and investors talk about NRR (they want the num­ber to go up). Either is fine; pick one and stick with it with­in a giv­en report. Detailed mechan­ics are in the net rev­enue reten­tion guide.

Should reac­ti­va­tion MRR be count­ed as expan­sion or as new MRR?

Most oper­a­tors bun­dle it into expan­sion. I pre­fer break­ing it out as a sep­a­rate line because reac­ti­va­tion behav­ior tells you some­thing dif­fer­ent from organ­ic expan­sion — it says either “the prod­uct is sticky enough that cus­tomers return after try­ing alter­na­tives” or “the sales team is win­ning back fix­able churns.” Both are use­ful, both are dif­fer­ent from “exist­ing cus­tomers buy­ing more.” If you bun­dle it, doc­u­ment the choice and apply it con­sis­tent­ly.

Does seat expan­sion at an exist­ing cus­tomer count as expan­sion MRR or new MRR?

Expan­sion MRR, in essen­tial­ly every frame­work that mat­ters. The cus­tomer rela­tion­ship is the same; the dol­lars are larg­er. The only edge case is when the expan­sion hap­pens via a fun­da­men­tal­ly new prod­uct line that the cus­tomer was not buy­ing before — some oper­a­tors argue this is “cross-sell new MRR” rather than expan­sion. The arith­metic does not care; the seg­men­ta­tion does. If you are try­ing to mea­sure how well your land-and-expand motion works, treat seat expan­sion as expan­sion. If you are try­ing to mea­sure how well your cross-sell motion works, break new-prod­uct cross-sells out sep­a­rate­ly.

What is the rela­tion­ship between MRR churn and LTV?

Churn is the denom­i­na­tor of the LTV cal­cu­la­tion. LTV = ARPA ÷ Month­ly Churn Rate, where the rel­e­vant churn rate is usu­al­ly gross MRR churn (or rev­enue churn) for the LTV-per-dol­lar inter­pre­ta­tion. A 1% improve­ment in month­ly MRR churn length­ens implied cus­tomer lifes­pan from 50 months to 100 months — which dou­bles the LTV. That is why reten­tion work com­pounds in val­u­a­tion impact: small improve­ments in the churn rate trans­late into large mul­ti­pli­ers in the LTV-to-CAC ratio that buy­ers care about. The math is detailed in the LTV/CAC guide.

The Bottom Line

MRR churn is the most impor­tant reten­tion met­ric a SaaS busi­ness reports. Get it right and you have a clean sig­nal that lets you sep­a­rate “we are grow­ing because we are buy­ing cus­tomers” from “we are grow­ing because our exist­ing cus­tomers are pay­ing us more every month.” Get it wrong — by mul­ti­ply­ing instead of com­pound­ing, by lump­ing new logos into expan­sion, by hid­ing con­trac­tion inside churn — and you pro­duce a num­ber that looks like a met­ric but func­tions like noise.

Three habits sep­a­rate the founders whose MRR churn num­bers I trust from the ones whose num­bers I have to recom­pute on a coach­ing call. They report gross and net side by side, every month, with the trail­ing six months vis­i­ble. They use the com­pound for­mu­la for annu­al con­ver­sions, every time. And they treat the num­ber as a man­aged lever — diag­nose, inter­vene in one place, mea­sure the inter­ven­tion against a spe­cif­ic cohort, scale what works — not as a quar­ter­ly aspi­ra­tion. The math com­pounds; the dis­ci­pline com­pounds hard­er.

If your num­ber is wrong this quar­ter, do not pan­ic and do not hire a cus­tomer suc­cess team to fix it. Pull the diag­nos­tic, find the upstream dri­ver, fix the struc­tur­al cause. Then watch the num­ber move. That is how SaaS eco­nom­ics work when they work.