Churn

Gross dollar churn has held near its floor across four surveys, six data years, and roughly 390 respondent-years. The current edition puts it at 14–15%. It has not spiked, and it has not improved. It has simply refused to move. That persistence is the structural fact beneath the entire report: the base rate of revenue loss is fixed at a level high enough to demand constant offsetting expansion, and expansion is exactly what has weakened.

One note on how that floor is quoted. The survey has restated these years, and not gently: KBCM-2024 put 2022 and 2023 gross retention at 89%, and KBCM-2025 restated them to 86% and 85%. Read as churn, the same two years moved from 11% and 11% to 14% and 15%. Every one of those restatements ran in the same direction, which is worth sitting with: the floor did not rise, the earlier estimate of it was too kind. This report quotes the KBCM-2025 vintage throughout and does not blend editions into a single range. The standing rule on restated series is explicit: never blend; cite the edition shown.

Frozen Churn

Gross dollar retention has sat near 86% for the length of the series. On its face, stability reads as health. It is not. A frozen 86% GDR means the median company loses roughly 14% of its recurring revenue every year to outright churn before any downsell, and it has made no progress against that number in six years of data. Stability here is not the absence of a problem. It is the absence of the one improvement that could offset softening net retention. With churn fixed, every point of NDR decline has to be paid for out of expansion, and there is less of that every year.

The Net Magic Number Stuck at 0.50

The Net Magic Number measures how much net new ARR a company generates for each dollar of sales and marketing. Across the market it has held at 0.50 for four consecutive years (2022–2025E). Every dollar of S&M returns fifty cents of net new ARR once churn and contraction are absorbed. The growth engine has not accelerated, decelerated, or reset. It has stalled at a fixed, mediocre level of efficiency.

The stall is more troubling than a decline would be. A declining number is a problem you can watch and respond to. A number frozen at 0.50 through four years of rising customer-success investment indicates that the additional spend is producing expansion revenue, not churn reduction, and that the underlying churn floor is simply not responding to more effort. Consider a company growing 30% on a $50M base: it needs $15M of net new ARR to hit plan, and at a Net Magic Number of 0.50 that requires roughly $30M of sales and marketing to produce. Half of what the machine generates is consumed before it reaches the top line. Calculated

Source: KBCM/Sapphire Survey 2022–2025Verified

The Churn Tax

The gap between the gross and net magic numbers has a name in this report: the churn tax. It is the share of sales-and-marketing output consumed by replacing revenue that churned rather than by adding revenue that grows the business. That tax runs 22–24% of S&M spend across the recent series, and the survey’s own 2025E estimates imply 29%. Close to a quarter of the entire sales-and-marketing budget, rising toward a third, buys nothing but the ground the company already stood on. It does not appear as a line item anywhere. It is diffused across bookings, renewal, and expansion, which is precisely why it persists.

Gross and Net Magic Numbers: KBCM-2025 (p9 and p34), incl. 2025E–2026E survey estimates; churn tax derived (gross − net)Calculated

Derived: Gross Magic − Net Magic, per yearCalculated

Churn by Contract Length

Contract length is the strongest single predictor of churn in the dataset, and it is the one a company fully controls. Month-to-month terms run roughly 14% annual churn; multi-year terms run roughly 3%. That is a 79% reduction from a single variable, holding product, pricing, and customer-success investment constant. No operational program in the data produces a comparably documented effect. The relationship is a gradient, not a switch: churn falls at every step from sub-annual to multi-year, which means even a partial shift in contract mix moves the number.

Source: KBCM/Sapphire Survey 2024 p12 (2023 data), churn by contract lengthVerified

Churn by Contract Size

Contract size shows the same defensive logic as length. In the survey’s 2018 data, median gross dollar churn fell monotonically across every ACV band: 25.9% for contracts under $5K, 12.9% at $5–25K, 12.3% at $25–50K, 11.7% at $50–100K, and 9.8% above $100K. The 2020 data repeats the gradient across deal-size bands, from 22% below $15K through 16% and 9% to 6% above $250K. In both years the smallest contracts churned at roughly two-and-a-half to nearly four times the rate of the largest, and no band broke the ordering. Contract size functions as a churn lever in this dataset, the way contract length does: a structural variable that sets the churn floor before any operational program runs.

Source: KBCM/Sapphire Survey 2019 p57 (2018 data); 2020 figures from KBCM-2020 p21, deal-size bandsVerified

The mechanism is not identical to length’s, though the effect looks the same. Larger contracts typically mean more stakeholders signed off on the purchase, more implementation and integration work sunk into the account, and dedicated CS coverage rather than pooled or self-serve support, all of which raise the cost and organizational friction of leaving, independent of whatever the renewal term happens to be. Inference

Read against that gradient, the industry’s drift upmarket looks like retention policy. Within the latest survey vintage, median ACV (measured as ARR divided by total customers) rose from $44K in 2022 to $49K in 2023 and $60K in 2024, a 36% increase over exactly the window in which net retention weakened. Nothing in the band data shows larger contracts buying faster growth. What they buy, everywhere the cut exists, is a structurally lower churn floor. The move upmarket is therefore best read as gross-retention defense: the market shifting toward the contract sizes that churn least while the go-to-market machine delivers less on the acquisition side. Inference

Two limits belong next to this finding. The by-ACV churn cut exists only for the 2018 through 2020 data years; the survey stopped publishing it, so no gradient has been measured in the efficiency era, and the figures above are pre-correction measurements applied to a post-correction question. The relationship is also correlational rather than causal: companies selling $100K contracts differ from companies selling $5K contracts in segment, buyer, and product maturity, and composition carries some unknown share of the gradient.

The Adoption Paradox

The market is pulling its strongest lever in reverse. Even with the 79% reduction well documented, multi-year adoption fell from 48% to 26% in a single survey year, while the share of companies on one-year-or-shorter terms rose from 52% to 74%. Companies understand the lever and are moving away from it, driven by buyer-side risk aversion in a volatile pricing environment: customers resist long commitments exactly when vendors most need the retention those commitments provide.

This is the clearest example of the report’s larger pattern. Contract structure is a capital-allocation decision, reversible and within the company’s control, and the market is making it in the direction that raises churn at the moment churn can least be afforded. The frozen churn floor is not immovable. It is being held in place by choices. Inference

Source: KBCM-2024 → KBCM-2025Verified