Renewal Rate: Why 97% of Contracts Renewing Can Still Cost You ARR
Renewal rate is the only retention metric whose denominator is a calendar of expiry dates — so a longer contract term can lift it without a single customer being happier. The formulas, the traps, and what CrowdStrike, Bynder, ServiceMax, Keeper Security and Movable Ink put on the record.
On this page
CrowdStrike’s enterprise customers renewed. Through the fiscal year that contained the 19 July 2024 outage, and again in the quarter after it, gross retention held at 97%. Dollar-based net retention still fell seven points, from 119% to 112%, and one of the reasons sits in a list inside the fiscal 2025 10-K. Among the concessions the company bundled into what it calls customer commitment packages is the line “subscription period extensions” — the same contract value, stretched over more months. That is a renewal. It counts as a renewal. It subtracts annualised revenue.
Renewal rate is the metric that records the first sentence of that paragraph and can miss the last one. It is also the number most enterprise SaaS companies actually run on, because it is the only retention figure shaped like the way their customers buy: not continuously, but once, on a date, for a term.
Renewal rate has a calendar for a denominator. Every other retention metric divides by a population — the customers you had, the dollars you had. Renewal rate divides by a schedule: the contracts whose term expired inside the window. Everyone else in the book is not retained, merely not yet asked. That one difference makes it the right number for a term business and the easiest retention figure in SaaS to move without changing anything a customer feels.
Logo renewal rate = contracts renewed ÷ contracts that came up for renewal in the periodOnly accounts whose term expired inside the window enter the fraction. A mid-term cancellation is not a renewal event at all.
Dollar renewal rate = renewed contract value ÷ contract value that came up for renewalComputed on annualised value, before any expansion sold away from the renewal table — which is what makes a stretched term visible.
What this metric owns, and what the neighbours own
Three related explainers sit beside this one, and they are not interchangeable. Renewal rate is the term-boundary view; it only exists if your customers sign terms.
The general case: the four choices — unit, period, cohort, denominator — sitting under any retention figure however you bill. Start there if your customers can leave on any day of the month.
The dollar ledger: gross against net, expansion added back, and the benchmark spread founders disclose. That guide owns the gross-versus-net argument, and this one does not repeat it.
Where all of them end up. In a term business the inversion changes shape — expected life comes out in terms, not months — which is a different arithmetic from the one the LTV guide takes apart.
One conversion the churn guide teaches does not carry across. Monthly churn compounds into an annual figure. A renewal rate compounds into nothing, because it is already denominated in terms: if your contracts run a year, the rate is annual by construction, and if they run three, annualising it is not arithmetic, it is a different population.
The quiet lift: longer terms
Because the denominator is a calendar, you can shrink it. Sell three-year contracts instead of one-year contracts and two-thirds of the book becomes ineligible to leave in any given year. On our arithmetic, a book entirely on three-year terms exposes about a third of itself to a renewal decision annually; two years’ worth of accumulated dissatisfaction stays invisible to the metric until its date comes up.
Founders describe this as a strategy rather than a measurement caveat. Malwarebytes CEO Marcin Kleczynski, in a February 2018 interview, walked through a consumer-to-enterprise motion — employees installing the consumer product on work machines until IT signs a contract — that produced roughly half of all enterprise deals, at upwards of $3,000 a year on a two- or three-year contract. GetLatka’s write-up of that tape puts the 2018 sales and marketing focus on B2B for exactly this reason: business sales generate much higher renewal rates. Keeper Security, under CEO Darren Guccione, recorded a 90–92% annual retention rate in 2021 and attributed it directly to a model that pushes annual subscriptions over monthly plans, committing customers for longer periods.
Both advantages are real. Longer commitments genuinely do reduce the number of chances a customer gets to leave. But a renewal rate that improved because the sales team started closing three-year deals last quarter is reporting a change in contract mix, and a renewal rate is only as good as the fraction of the book it describes.
10 termsexpected customer life implied by inverting a 90% renewal rate — 1 ÷ (1 − 0.90), our arithmetic on Keeper Security’s disclosed 90–92%
Nobody should publish that number. It is the constant-hazard assumption in term clothing: it assumes the eleventh renewal is as likely as the first, and a decade of expected life is what the assumption is worth. The inversion is useful only inside a single cohort with a single term length.
A renewal is not a yes or a no
The renewal event is a negotiation, and negotiations have more than two outcomes. CrowdStrike’s 10-K lists what the company put on the table after the outage:
- Discounting — a straight price concession on renewal.
- Additional modules — more of the platform at no incremental ARR.
- Professional services — delivery cost absorbed by the vendor.
- Flexible payment terms — cash timing traded for goodwill.
- Subscription period extensions — the same contract value stretched over more months.
All five are a yes in the logo renewal rate. Four cost dollars a dollar renewal rate would catch. The fifth is the one most operators would never have modelled: extending a term without raising the total spreads the same money across more months, so the annualised contract value falls while nothing at all was lost. The filing says so in as many words — the packages “have resulted, and are expected to continue to result, in increased contraction, due to elongated subscription terms, and decreased upsell dollar values.” A concession that cost no cash cost real points of net retention, and kept costing them for the length of the extension.
Which is the case for publishing the dollar version beside the logo version rather than instead of it. The logo rate tells you whether the relationship survived. The dollar rate tells you what the relationship is now worth per year, and those two answers diverge most sharply in precisely the quarters you most want to understand.
It arrives up to a full term late
Renewal rate is the most lagging indicator in the retention stack. A customer who gave up in month two does not register until month twelve; on a three-year term, not until month thirty-six. The number is a receipt, not a warning.
The operators who handle this well watch consumption and treat the renewal as confirmation. Jumio, in a July 2017 interview, ran contracts from a $30,000 floor to well over $1 million a year in which customers prepaid a year’s worth of identity verifications and burned down the inventory. CEO Stephen Stuut’s single operating metric was verifications processed — which is the burn-down curve, and the burn-down curve is the renewal decision showing its hand eleven months early. An account halfway through its year with three-quarters of its block unused has already answered.
The boundary is also something you create rather than something you have. vFairs sold virtual-event software as a discrete single-event licence before chief revenue officer Michael Burns repositioned the identical product as the platform an organisation runs all of its events on. The software did not change. The metric came into existence.
A discrete single-event licence sold on features, more service engagement than software subscription — and renewed one event at a time, if at all. No term boundary, so no renewal rate to report.
The same software framed as how an organisation runs all of its events, sold into the enterprise on annual and multi-year contracts. In Burns’s phrase, “proper SaaS” — and a calendar of expiry dates to measure against.
On the record
Almost nobody publishes a renewal rate
Search the GetLatka archive for a founder-stated renewal rate and you will struggle to find one. What founders publish is gross retention — which, in a book where every contract runs twelve months, is the dollar renewal rate wearing a different name, and stops being it the moment the terms stop matching.
| Company | Gross retention disclosed | Net retention | Contract shape |
|---|---|---|---|
| Bynder (early 2018) | 93% gross revenue retention | 104% | Annual, ACVs $30K to nearly $1M |
| Movable Ink (mid-2018) | Declined; “pretty average” | North of 110% | Annual, some multi-year, $30K floor |
| Keeper Security (2021) | 90–92% retention, unit unstated | — | Annual subscriptions over monthly plans |
| ServiceMax (Nov 2022) | Mid to high 90s | 115–121% | Enterprise ARR, $150M+ disclosed |
| CrowdStrike (FY2025) | 97% | 112%, from 119% | Multi-year subscription terms |
Founder disclosures from the GetLatka archive, plus CrowdStrike’s filings, each carrying the contract shape its source attached. None of these figures is labelled a renewal rate; in an all-annual book the second column is one.
Read the last column and the second column together and the spread mostly explains itself. The tight band from 90% to 97% is not five companies with comparable products; it is five different contract calendars, and Bynder’s 93% on twelve-month terms is a harder number than a mid-90s figure earned on multi-year ones.
Reporting a renewal rate that survives diligence
- Publish the calendar with the rate. Contract value coming up for renewal in each of the next four quarters, next to the rate achieved on the last four. Without it, the number describes an unknown fraction of the business.
- Report dollars beside logos. The logo version says whether accounts stayed. The dollar version, on annualised value, is the only one that catches a renewal negotiated down — or stretched out.
- Hold the term constant, or disclose the mix. If weighted average contract length moved, say by how much. A rate that improved alongside a shift from one-year to three-year deals has said nothing about the product yet.
- Keep renewal and expansion apart. Upsell agreed at the renewal table is not evidence the renewal was healthy; it is the most likely thing to be hiding that it was not.
- Count a partial as a partial. A customer renewing at 60% of last year’s value is 100% of a logo and 60% of a dollar. Rounding that to a yes is how a book shrinks through four consecutive quarters of perfect renewal rates.
Vivek Sharma had most of this in one answer. Movable Ink was doing $40 million of ARR on $14 million raised when he sat down with Nathan Latka in mid-2018, selling 500 consumer brands on annual and some multi-year contracts from a $30,000 floor. Asked about churn, he refused the frame before answering it: “We are an ARR business, not an MRR business,” he said, so “you can’t really look at churn on a month over month basis because it’s not being intellectually honest.” What his team looks at instead is the cohort that landed last August, up for renewal now, and how many of them stayed. He would not publish the gross figure, calling it pretty average, while putting net revenue retention north of 110% — and he was exact about why the two have to be read apart.
“You’ve got to treat both of those independently, otherwise you can mask a very leaky bucket.”
Vivek Sharma, co-founder and CEO, Movable Ink
SourcesFounder disclosures from the GetLatka archive: Movable Ink (Vivek Sharma, interview recorded mid-2018), Malwarebytes (Marcin Kleczynski, February 2018), Jumio (Stephen Stuut, July 2017), vFairs (Michael Burns), Bynder (Chris Hall, shortly after the early-2018 Webdam acquisition), Keeper Security (Darren Guccione, 2021) and ServiceMax (November 2022). CrowdStrike figures from its Form 10-K for fiscal 2025, its Q4 and full-year fiscal 2025 results release of 4 March 2025, and its Q1 fiscal 2026 results release of 3 June 2025. The three-year-term denominator estimate and the inverted 1 ÷ (1 − renewal rate) expected-life figure are our arithmetic applied to the disclosed rates.
