Growth Endurance
When you need this method
Growth is easing off and nobody in the room can say whether that is normal. The board sees 43 percent after 75 and reads it as a sales failure. The team reads it as the base having grown. Both sides argue without an expected value, so the louder voice wins. As long as there is no yardstick for ordinary growth decay, every slowdown is either dramatized or explained away, and scarce resources go to the wrong place.
Approach
- 1Collect year-end ARR for the past three to four years and derive the growth rate for each year.
- 2Compute endurance for every pair of years: this year's growth rate divided by last year's growth rate.
- 3Compare the result against the published bands: around 70 percent counts as normal for privately held companies, 75 percent as better, 80 percent as strong; listed companies historically sat near 80 percent.
- 4If your value falls clearly below the band, stop explaining the slowdown with the base effect and go into the causes: an exhausted core segment, weakening retention, missing sales capacity, price structure.
- 5For planning, apply your own historically measured endurance to the current growth rate and extend the ARR path several years out instead of modelling a wished-for rate.
- 6State the gap between planned growth and endurance-implied growth, and name it for what it is: a bet that needs its own justification.
Typical application
A typical B2B SaaS in the HR space grows from 8 to 14 to 20 million euros ARR over three years, so 75 percent and then 43 percent. The board reads the drop as a sales problem and asks for more headcount. Endurance comes out at 57 percent, clearly below the usual band, so the slowdown is not just the base effect. Looking into the causes shows two things: the core segment is largely sold through, and gross retention fell in the prior year. Additional sales headcount would not have changed either. The budget goes into a second segment and into retention instead, and the multi-year plan is rebuilt on an endurance-based path.
Limits and counter-indications
The stability of endurance is a population finding from venture capital portfolio data, not a law for the individual firm. Research on firm growth finds only weak persistence at the level of a single company, and even negative autocorrelation among the fastest growers, where sustained high growth is the exception. Using the metric as a forecast for one company rather than as an expected value overstretches it. The bands also come from a particular market phase; more recent practitioner data reports a markedly lower figure. At low base rates the ratio loses meaning, since 80 percent endurance on 3 percent growth means something entirely different than on 60 percent. Acquisitions, pricing changes, and single large deals distort the value, and with fewer than three clean year-end figures it cannot be computed at all.
How to measure impact
Track endurance for each pair of years over at least three years and pair it with GRR and NRR, since retention and expansion are the most direct levers on the value.
Related methods
Sources
- 1.Scaling to $100 Million, Bessemer Venture Partners (BVP Atlas), Abschnitt Growth Endurance (opens in a new tab) · Bessemer Venture Partners · 2021 · practitioner source · describes the methodCarries the definition and the procedure: growth endurance is the share of a growth rate retained from one year to the next. In the private cloud market growth decays by roughly 30 percent per year, so the next year sits at about 70 percent of the current one; in the BVP Nasdaq Emerging Cloud Index endurance runs near 80 percent. The source also shows the application as an extrapolation toward an ARR target. Limit: this is a heuristic drawn from one investor's portfolio data, with no disclosed sample size and no peer review.
- 2.Understanding Revenue Growth Endurance, Mostly Metrics (CJ Gustafson) (opens in a new tab) · Mostly Metrics · 2025 · practitioner source · describes the methodCarries the explicit calculation and the reading rules: this year's growth divided by last year's growth, in the worked example 40 percent divided by 50 percent equals 80 percent endurance. For privately held companies the source cites 70 percent as good, 75 percent as better, 80 percent as best; below 100 percent means deceleration, above 100 percent acceleration. It names the key limit itself: the same endurance value means something different on 3 percent growth than on 20 percent. Limit: a secondary account with no data collection of its own, the bands come from the Bessemer work.
- 3.2025 SaaS Performance Metrics Benchmarks, Benchmarkit (opens in a new tab) · Benchmarkit · 2025 · practitioner source · provides benchmark figuresSupplies a current data point on the level of the bands: the reference value for growth endurance used to sit near 80 percent and has fallen to roughly 65 percent over the past two years. This shows the bands depend on the market phase and must not be read as a fixed constant. Limit: the source carries growth endurance only as an explanatory side note to its growth section, with no sample size and no methodology specific to this metric.
- 4.Coad, Daunfeldt, Halvarsson: Firm Age and Growth Persistence, Konferenzpapier Innovation Forum VI, Paris (opens in a new tab) · Innovation Forum VI (University of Paris Ouest Nanterre) · 2014 · academic and scholarly literature · limits the methodSets the empirical boundary for the metric. Drawing on all Swedish limited liability firms from 1997 to 2010, the authors show that autocorrelation of sales growth is positive for young firms but turns increasingly negative with age, and that persistence becomes negligible after fewer than ten years. They also summarize the wider evidence that the strongest negative autocorrelation appears precisely among the ten percent fastest growing firms, making sustained high growth a very unlikely path. For the method this means a stable endurance factor is an aggregate finding, not a licence to extrapolate a single company. Limit: the study covers the entire Swedish economy rather than cloud business models, so it does not refute the bands but bounds their reach.
Origin: Bessemer / D'Onofrio