CAC Trap

Acquisition economics have deteriorated into a trap: fully-loaded CAC runs 2.8× reported CAC, new-only payback has stretched to 37 months, and roughly 37% of new customers churn before they ever pay back. The trap is that the true cost was always hidden, and the mechanism that hid it, net-retention expansion quietly subsidizing acquisition, is the one this report shows weakening.

Fully-Loaded CAC

Reported CAC, sales-and-marketing spend divided by new customers, badly understates true cost. Load in the R&D, onboarding, customer-success, and overhead that winning and standing up a new account actually consumes, and fully-loaded CAC runs about 2.8× higher across every company size.

Here is what that gap means in practice. A company divides its S&M line by the logos it closed and reports a CAC that looks affordable. The number is real, but it is partial: the engineers who built the features that closed the deal, the onboarding team that made the account stick, the support that carried it through month three, and the overhead behind all of it never entered the calculation. Fold those in and the true cost is 2.8× what was reported. The difference did not vanish. It was paid by the existing customer base, whose revenue quietly funds the acquisition machine. That is the net-retention subsidy, and it holds only as long as net retention does. Calculated

One thing the 2.8× is not: a measurement. It is a modeling construct, and the allocation choices inside it should be visible before the number is repeated. Folding onboarding and the acquisition-adjacent share of customer success into CAC is conventional; allocating R&D to new-customer acquisition is the aggressive step, because product investment serves the entire base, and a different analyst making a defensible different allocation would land somewhere other than 2.8×. The construct’s value is directional and it does not depend on the second decimal: on any reasonable allocation, reported S&M-only CAC understates the all-in cost of standing up a new account by a wide multiple, and the gap is funded by the existing base. Quote the 2.8× with its construction stated, or quote the direction without the number. Inference

Report-derived composite: reported S&M CAC with R&D, onboarding, customer-success, and overhead allocated to new-customer acquisition added in; distinct from KBCM’s S&M-only efficiency-ratio sense of the same phrase. See On “Fully-Loaded CAC” for the full disambiguation.Calculated

Payback Deterioration

Fully-loaded payback held at 24 months, but new-only payback stretched from 31 to 37 months, a widening gap (7 → 13 months). New customers now carry more of the cost upfront, while the net expansion uplift that would offset it has thinned (NDR 109→101).

Derived: CAC payback by cohort, 2022 vs 2024Calculated

Read honestly, the flat blended line is the strongest counter-evidence in this chapter’s own data, and it should be said plainly rather than left for a critic to find: fully-loaded payback holding at 24 months through the entire correction is, on its face, the healthiest series in the dataset. What decides between the two readings is what held it flat. The blended figure nets the expansion of the existing base against the cost of acquiring the new one, so a flat blend with a stretching new-only line does not show acquisition economics holding; it shows the subsidy still working. New customers got 6 months more expensive to pay back, and the existing base absorbed the difference. That is stability the way a co-signed loan is creditworthiness: real until the co-signer’s capacity thins, and the co-signer here is the net retention this report documents weakening. If the blended line is still flat two editions after coverage has run at 1.07×, that reading needs revisiting; the more likely sequence is that the blend follows the subsidy down, late, the way blended metrics always report last. Inference

Churn Before Payback

At 37-month payback and 14% annual churn, roughly 37% of new customers churn before they pay back. That is more than a third of acquisition spend lost outright, and the arithmetic behind it is worth showing, because seeing it is what makes the figure hard to argue with. Thirty-seven months is about 3.1 years. A cohort losing 14% a year keeps 0.86 of itself each year, so after 3.1 years it keeps 0.86 to the 3.1 power, or roughly 63%. The missing 37% left before the money they cost was ever recovered. Healthy SaaS companies, the Rule-of-40 achievers, lose under 10% this way.

Derived: 14% annual churn (100% − 86% GDR) compounded across the 37-month payback; 0.86^3.1 ≈ 63% retained, so ≈ 37% lost.Calculated

Subsidy Erosion

The scissors close: rising acquisition cost meets declining NDR. As net retention softened (109%→101%), the net-expansion uplift that subsidizes acquisition thinned. New customers now carry more of the cost upfront with less expansion to offset it. (Expansion’s share of new ARR rose over this period. See The Expansion Myth; what changed here is net retention, not the expansion mix.)

Derived from KBCM/Sapphire Survey 2022–2025 NDR trajectory + CAC-payback cohortsCalculated

Read against the report’s structure, the CAC trap is not a separate problem so much as the price of the others. Acquisition looked affordable for a decade because expansion and retention silently subsidized it; the reported CAC was only ever the visible portion of the real cost. As net retention falls, that subsidy thins and the full 2.8× cost surfaces onto the income statement. A company that cut R&D and shortened contracts to protect margin then finds acquisition more expensive precisely because those moves weakened the retention that made acquisition cheap. Inference