The 94% Problem: Rebuilding NRR Past 110%

Derek SimmonsNinety-four percent, median net revenue retention for companies under five million in ARR. The typical tiny SaaS shrinks by six points a year without even trying.

Elena ReyesOkay, but wait for it. Same neighborhood, different report, and the ninetieth percentile is near a hundred and eighteen.

Derek SimmonsSame neighborhood?

Elena ReyesClose enough to make you squint.

Derek SimmonsAh, squinting. Noted. I'm Derek, that's Elena, and this is ARR Autopsy.

Elena ReyesBenchmarkit's twenty twenty-six report is the ninety-four. SaaS Capital's bootstrapped companies, three to twenty million, are the hundred and eighteen at the top end.

Derek SimmonsSo which is it? Is the gap an error somebody made in a spreadsheet, or is it money sitting on the table?

Elena ReyesOh, I'm betting on money, real money, the kind you can spend.

Derek SimmonsSure you are. Quick housekeeping. No guest today, and nobody sent us a founder spreadsheet.

Elena ReyesRight. So the hundred customer base we build is a hypothetical, ours, cooked up in this room.

Derek SimmonsWhich means every dollar figure we invent gets labeled as invented. Nothing made up gets to wear a source's name.

Elena ReyesGood rule. If it comes from a report, we say which report. If it's ours, we say it's ours.

Derek SimmonsAnd the tool for sorting out whether you've got a churn problem or an expansion problem is gross retention against net retention. We'll get there.

Elena ReyesPatience, Derek.

Derek SimmonsPatience. Never. But before we do math on those numbers, we have to know what each one actually measured.

Elena ReyesBecause a headline number can hide a lot.

Derek SimmonsOkay, first problem. The brief says SaaS Capital's median for bootstrapped three to twenty million dollar companies is a hundred and three percent. Two advisory write-ups citing SaaS Capital say the twenty twenty-five survey says a hundred and four percent.

Elena ReyesOne point. Who cares?

Derek SimmonsI care, because I can't tell you which one is on the SaaS Capital page. We didn't confirm it in our materials.

Elena ReyesWow, a podcast that admits it didn't check.

Derek SimmonsDeadpan growth strategy. Now, the hundred and eighteen percent at the ninetieth percentile shows up in both write-ups, so that part holds.

Elena ReyesSo is the one oh three a typo?

Derek SimmonsMaybe not. It could be a different survey year, same shop, different edition, and the number moved a point. I'd bet on that over a mistake, but it's a bet.

Elena ReyesFair. Now, the bigger gap. These two benchmarks aren't even measuring the same crowd.

Derek SimmonsWalk me through it.

Elena ReyesBenchmarkit covers companies under five million in ARR. SaaS Capital covers bootstrap companies between three and twenty million, different size band, and one is filtered by how they got funded.

Derek SimmonsSo a company at four million ARR could sit in both samples or in one.

Elena ReyesExactly, and it gets muddier. An advisory post from Levera Partners puts companies at one to ten million in ARR near ninety-eight percent.

Derek SimmonsWhich sits between the two benchmarks.

Elena ReyesCute. Then another advisory post, Fiscal Lion, puts SMB-focused companies contract value under twenty-five thousand at about ninety-seven percent.

Derek SimmonsSo the smaller and cheaper your customer, the lower the number. Classic problem.

Elena ReyesCareful, though. Two posts isn't a law. It's a pattern across two advisory write-ups, and each has a sample we can't see.

Derek SimmonsRight. We don't know how either one picked its companies, so I'm not gonna pretend the ordering is proven.

Elena ReyesOh, and there's a fourth. Alephs' page reports a median across all SaaS for a hundred and two percent labeled twenty twenty-five from three hundred and forty-two companies.

Derek SimmonsHold on. That's a median for everybody, and it's above Benchmarkit's number for small companies?

Elena ReyesBy a lot. And the same page splits by pricing model, usage-based at a hundred and eight percent, seat-based at ninety-eight percent.

Derek SimmonsTen points from how you charge. That's bigger than some of the gaps we've been arguing about.

Elena ReyesKeep that in your pocket, because it comes back when we talk levers.

Derek SimmonsNoted. So the picture is a pile of medians from different samples, different years, different definitions of small.

Elena ReyesAnd here's the practical bit. A median is the middle company. Half the companies in the sample did better and half did worse, so none of these numbers is a promise.

Derek SimmonsSo when somebody says the benchmark is ninety-four, the right reply is, "Whose ninety-four?"

Elena ReyesAnd none of them is your company, your customers, and your actual churn.

Derek SimmonsPlayfully, but correct. Okay, gap understood. Now, what does a hundred and ten percent actually mean for a hundred customers? Okay, so a hundred customers. Say each pays the same, and I'm making that up. We're not citing anything here. Call it a thousand dollars a month apiece. Hundred thousand a month all in.

Elena ReyesHold on. Same price for everybody? Nobody has that. But fine, it's a toy. Keep going.

Derek SimmonsToy is the point. Levera Partners, an advisory firm, makes a simple observation. At a hundred and ten percent NRR, your installed base grows ten percent a year with zero new logos.

Elena ReyesZero. Nobody signs, and you still grow.

Derek SimmonsRight. So on my toy base, that's ten thousand dollars a month of new revenue coming only from people already paying you.

Elena ReyesOkay, but that's the destination. You're starting at ninety-four, the Benchmarkit number. How far is the climb?

Derek SimmonsSixteen points. On a hundred thousand dollars a month, that's sixteen thousand dollars of net gain you have to find.

Elena ReyesNet meaning after you lose.

Derek SimmonsMeaning after the losses. Hold that thought. First, the pure expansion version. Say twenty customers upgrade, and I'm inventing this. Each adds eight hundred dollars a month.

Elena ReyesTwenty out of a hundred, one in five of your customers saying yes to a bigger plan sounds doable on a whiteboard.

Derek SimmonsOn a whiteboard, everything's doable. Which twenty? Are those the happy ones who were going to stay anyway?

Elena ReyesProbably. Plot twist, that's usually who upgrades.

Derek SimmonsSo the expansion comes from the healthy end, and the unhappy end is still leaking. Fine. Now, how big are each bump? Eight hundred on a thousand-dollar account is nearly doubling.

Elena ReyesOoh, fair. Loosen it. Forty customers, four hundred each. Same sixteen thousand, but now it's forty percent of the base moving and a smaller ask per account.

Derek SimmonsThat's a different company. One is twenty big sales conversations, the other is a pricing page nudge forty times.

Elena ReyesAnd which is why the average hides everything. Same sixteen-point lift, two totally different jobs for your team.

Derek SimmonsAnd notice the percentage can't tell those two apart.

Elena ReyesNope. It only knows the total.

Derek SimmonsNow the ugly version. What if two customers do the whole thing? Two accounts each add eight thousand a month.

Elena ReyesSixteen thousand. Ninety-eight customers did nothing, and the spreadsheet still says you hit your number.

Derek SimmonsAnd if one of those two walks next year, then half your expansion vanishes. Net gain drops to about eight thousand, so on the toy base, you're roughly a hundred and two percent before any other losses, not a hundred and ten, and no dramatic event needed.

Elena ReyesHmm. So the headline looked healthy, but it was resting on two accounts.

Derek SimmonsAnd it gets worse because I've been quietly assuming nobody left. Every one of those scenarios treats the one hundred as a fixed crowd. Real bases lose people while others upgrade. So put some losses in the toy. Say five customers cancel a thousand a month each. Five thousand a month walks out the door. Now, to end the year at one ten, the upgrades don't have to find ten thousand, they have to find fifteen. Half as much expansion for the same target. And if the people leaving are the bigger accounts, the math moves again, which makes me suspicious. Two companies could print the exact same net number, one losing five customers and the other losing none.

Elena ReyesSo how would you even tell them apart?

Derek SimmonsYou look at gross retention, which define it for me like I've never met a spreadsheet.

Elena ReyesGross revenue retention counts losses only, churn and downgrades. It's capped at hundred percent because nothing you upsell can push it higher.

Derek SimmonsSo it can't be flattered.

Elena ReyesNever. NRR is the one that includes upsell and cross-sell, so we can be flattered all day.

Derek SimmonsWhich type of flattering are we worried about?

Elena ReyesThe kind where a few big upgrades cover a lot of cancellations. An advisory writeup of SaaS Capital survey treats GRR above ninety percent paired with NRR above a hundred and ten percent as the combination buyers want.

Derek SimmonsBoth at once?

Elena ReyesGreedy. Buyers are usually.

Derek SimmonsHang on. Is that what buyers actually do or what one write-up says buyers want?

Elena ReyesFair. It's the write-up's framing. It doesn't show us a stack of deals, so I'd take it as a picture of what a healthy pair looks like, not a rule.

Derek SimmonsFine, but is ninety percent even normal? Because I'm guessing the benchmark crowd has an opinion.

Elena ReyesOh, they do. The SaaS CFO analyzed the twenty twenty-six Benchmarkit report and says median GRR fell from eighty-eight percent to eighty-four percent. Fell. Down. And the seventy-fifth percentile fell from ninety-five percent to ninety-one percent, so even the good companies are leaking more.

Derek SimmonsSo the tide's going out and everyone's pointing at their expansion.

Elena ReyesThat's the risk the analysis flags. When GRR is falling, expansion can be hiding a churn problem. NRR looks fine, and underneath, you're refilling a bucket with a hole in it.

Derek SimmonsPut that on our toy hundred. What does the hiding look like?

Elena ReyesThe one that loses five customers shows a GRR of ninety-five, even if upgrades push its net number to match the one that lost nobody.

Derek SimmonsSame headline and only one tab shows the five who left.

Elena ReyesRight. Now, there's another way a headline flatters you, and that's one-off events. An advisory post from Fiscal Lion describes a nine million dollar ARR company at a hundred and eighteen percent NRR.

Derek SimmonsNice number.

Elena ReyesTwo accounts supplied sixty-one percent of the expansion. Both were one-time events. Strip them out, and NRR was ninety-seven percent.

Derek SimmonsHold on, from one eighteen to ninety-seven?

Elena ReyesYep, and I'll flag it. The post doesn't verify it. Treat it as a warning, not a data point. But it's a warning that gets louder when your base is small.

Derek SimmonsSo it's not only how high the number is, it's whether it repeats.

Elena ReyesRight. A one-time event doesn't come back next year, so it can't be counted on to hold the number up.

Derek SimmonsHow would I check that on my own hundred?

Elena ReyesDivide your accounts by expansion dollars. Add up the top three and divide by total expansion. If that share is most of it, your headline is really three customers.

Derek SimmonsSo a headline NRR, the benchmarks or yours, tells me basically nothing yet.

Elena ReyesNothing until you do two checks. Split out GRR, then look at how concentrated the expansion is.

Derek SimmonsAnd the split decides what you do on Monday.

Elena ReyesIf GRR is weak, fix churn first. Expansion on a leaky base is bailing with a teaspoon. If GRR is solid and NRR is flat, you have an expansion problem.

Derek SimmonsWhat counts as weak, though? Give me a line.

Elena ReyesThe sources give us pointers, not a line. The SaaS CFO's analysis has the median at eighty-four percent now, and the M&A write-up wants above ninety percent. In between is a gray zone, so read your own number against your own size band.

Derek SimmonsAnd if it's gray, look at the direction.

Elena ReyesExactly. A GRR that keeps sliding tells you more than one snapshot, the way the drop from eighty-eight to eighty-four tells you more than the eighty-four alone.

Derek SimmonsWhich is the better problem to have, honestly.

Elena ReyesIt's the more fun one. Nobody throws a party for a retention fix.

Derek SimmonsI'd attend. But say your GRR is healthy and expansion is the gap. Which levers do the sources actually name? Okay, so the sources named levers. Not a long list, and I wanna say that up front.

Elena ReyesOoh, managing expectations. Go on.

Derek SimmonsThe M&A advisory write-up, the one that explained GRR for us, recommends three things: usage-based pricing or hybrid, upgrade triggers at tier limits, and dunning fixes.

Elena ReyesDunning, the least sexy word in SaaS.

Derek SimmonsIt's failed payment recovery. Somebody's card expires, you send emails, and either they update it or they quietly churn.

Elena ReyesWhich is a GRR lever, not an expansion lever. Interesting that it's on the list.

Derek SimmonsRight. It plugs a leak, and the tier limit trigger is the opposite. A customer hits the cap on their plan, and the product itself asks them to move up.

Elena ReyesSo nobody has to book a call. Oh, I like that for a small team.

Derek SimmonsHold on, though. That's me reasoning. The write-up just lists it. It doesn't walk through a company doing it.

Elena ReyesFair. Noted. What about the pricing piece?

Derek SimmonsA second source, Fiscal Lions advisory post, is blunter. It says an SMB SaaS typically needs usage-based pricing to get above a hundred and ten percent.

Elena ReyesTypically. That's a soft word.

Derek SimmonsIt is, and it's the same post with the nine million dollar anecdote we already flagged as unverified, so hold it loosely.

Elena ReyesOkay, but why would usage pricing even do that mechanically?

Derek SimmonsBecause the bill rises when the customer's own volume rises. Nobody has to be sold anything, so expansion happens without a sales conversation.

Elena ReyesAgain, your reasoning. The post says what it says.

Derek SimmonsYou're enforcing the rule better than I am.

Elena ReyesSomebody has to. Is there any number on hybrid? One.

Derek SimmonsAn aggregator called Konabayev cites Benchmarkit's CY-24 report, so that's the prior year, and it gives hybrid subscription plus usage a median NRR of a hundred and ten percent.

Elena ReyesWait, median? So the target we've been chasing is just what the middle hybrid company does?

Derek SimmonsAccording to that aggregator, yes. But it's a secondary page quoting last year's report, and nobody checked it against the original.

Elena ReyesSo I wouldn't say it on a sales call. Wouldn't even say it to my cat.

Derek SimmonsUse it as a direction, not a benchmark.

Elena ReyesWhat would make you trust it more?

Derek SimmonsFinding the original Benchmarkit report and reading how it defines hybrid and who was in the sample. Until then, it's a pointer.

Elena ReyesOkay, now the uncomfortable part. The brief promised founder motions: upsell, cross-sell, the Tuesday morning details.

Derek SimmonsAnd we didn't get one. No founder spreadsheet, no sourced case study of a specific upsell or cross-sell motion. Zero.

Elena ReyesWhich is a little embarrassing given the title.

Derek SimmonsIt is. What we have is levers named by advisory posts and one prior year median. That's the honest inventory.

Elena ReyesAnd headcount. How many customer success people does it take? Sources don't say. No team sizes, no accounts per person ratio. Anyone who hands you one here is making it up. So it's an open question.

Derek SimmonsOpen and testable. Pull the accounts that expanded last year and check who touched them. Was it a CS person, or did the product do it?

Elena ReyesOh, that's good. If the tier limit accounts upgraded with zero human contact, you've learned something no benchmark can tell you.

Derek SimmonsAnd if every upgrade had a person behind it, you've learned the opposite.

Elena ReyesAnd I'd add a second cut. Of the accounts that hit a tier limit, how many upgraded and how many just sat there at the cap?

Derek SimmonsRight, because a trigger that fires and gets ignored isn't a lever yet. It's a notification.

Elena ReyesSame for dunning. Count the failed payments from last year and see how many you got back versus how many quietly left.

Derek SimmonsNone of that needs a benchmark. It's your own billing data and an afternoon with a spreadsheet.

Elena ReyesOkay, so we've got the levers, the caveats, and a homework assignment that costs nothing but an afternoon and a billing export. Now what do I actually do with my own dashboard?

Derek SimmonsLast piece, and it's the one that matters: turning all this into a decision. Okay, checklist time. Step one, compute GRR and NRR separately. Two numbers, two spreadsheet cells. Takes ten minutes, maybe fifteen with coffee.

Elena ReyesTen minutes? Founders will find a way to make it a quarter-long project.

Derek SimmonsThey'll build a dashboard for it. Step two, look at your top two or three expanding accounts and see how much of the expansion they supply.

Elena ReyesAnd if it's most of it?

Derek SimmonsThen your NRR is a rumor.

Elena ReyesOoh. Okay, and step three is the benchmark, but the right one.

Derek SimmonsRight. Pick the band that matches your size and your funding model, bootstrapped or venture backed, small or scale stage, not whichever number flatters you.

Elena ReyesAnd be honest about which band you're in. A company at four million ARR can fit more than one of the samples we talked about.

Derek SimmonsWhich is why the benchmark you pick should be one you can defend in a single sentence.

Elena ReyesWhich brings me to a reference point for the GRR check, because people always ask, "What's a good one?"

Derek SimmonsWhich type?

Elena ReyesScale stage bootstrapped. An advisory post from Levera Partners citing SaaS Capital's twenty twenty-five data puts the median GRR at about ninety-two percent.

Derek SimmonsSecondary source, so treat it as a pointer.

Elena ReyesSure, but the top performers in that same post sit at ninety-eight percent. That's a company losing almost nothing.

Derek SimmonsHuh. So if you're at, say, eighty-five, you're not shopping for upsell ideas.

Elena ReyesYou're shopping for a bucket.

Derek SimmonsOkay, that one's mine. I'll allow it. Step four and only step four, you choose.

Elena ReyesLeaky GRR, churn fix. Healthy GRR and flat NRR, expansion motion.

Derek SimmonsOrder matters.

Elena ReyesStop the leak before you build the ladder. A ladder on a sinking boat is just a fancier way to drown.

Derek SimmonsOh, wait, I like that more than I expected to. That's the autopsy. Here's what to do with it. Go open the spreadsheet tonight and write down your two numbers before you write down a single plan. Short version, since you've earned it, check GRR first, then ask whether the NRR comes from lots of accounts or from a couple of big ones.

Elena ReyesTwo questions, one spreadsheet tab. Your Tuesday just got easier.

Derek SimmonsMy Tuesdays are never easy, but sure.

Elena ReyesOkay, the ask. If you know a founder who's staring at a retention number and nodding like it means something, send them this episode.

Derek SimmonsJust one founder. Pick the one who says, "We're fine," a little too fast.

Elena ReyesSubscribe on YouTube or in whatever podcast app you use, and leave a review.

Derek SimmonsReviews are how this show keeps getting founders to hand over real numbers instead of polite ones.

Elena ReyesRight, and send us the spreadsheet if you've got one.

Derek SimmonsEven the ugly tab, especially the ugly tab.

Elena ReyesBecause our one hundred customers were made up and yours aren't. Yours are messier and a lot more interesting.

Derek SimmonsMessier, and they'll tell you whether you've got a leak or a ladder problem. Ours only taught us how to ask.

Elena ReyesThis was ARR Autopsy. Go split your GRR out.