The Invisible Function

The function that owns retention has no line of its own in the industry’s measurement system. The KBCM/Sapphire survey tracks dozens of spend variants for sales and marketing, R&D, and G&A; across all seven editions it has never once measured Customer Success spend as a share of revenue. CS cost appears only as an allocation, a record of where the cost is hidden, not of how much is spent. This deep dive follows the consequences: an expense folded into S&M just as S&M became the target of the deepest cuts, and a workload that grew 47% per CSM while net retention fell.

Never Measured

The absence is not a rounding error in survey design; it is systematic, and it shows in the questions themselves. For the sales side of the revenue organization, the survey asks, in one form or another edition after edition, what quota an AE carries, what share of it gets attained, how much new ARR a rep produces, how many sellers of each type a company employs, how they split between inbound and outbound, what they are paid in commission, and how long a deal takes to close. For Customer Success it has asked two operating questions in its history, how many accounts a CSM covers and how much revenue a CSM manages, and asked them only from the 2023 edition onward. CSM headcount appears once. The only other CS question is an accounting one: which expense bucket the cost goes into. The survey interrogates the half of the organization that wins revenue and takes the half that keeps it largely on faith, even though CSMs are 32% of the sales-org headcount in its own data. One detail is emblematic: the 2023 edition printed its CSM accounts-and-book table under the heading “Account Executive.”

Derived: question coverage compared across KBCM/Sapphire survey editions 2019–2025; CSM headcount mix from KBCM/Sapphire Survey 2025, p. 29Calculated

Where CS Cost Hides

When the survey does touch CS economics, it asks only where the cost is booked. The answer is: mostly inside sales and marketing. Across the four years the split was measured, an average of 57–68% of CS cost sat in S&M, with the S&M share drifting up and the COGS share falling from 36% to 25%. Support is the control case: its allocation converged to COGS long ago and stayed there. CS is the only major cost center in the survey without an accounting home.

Source: KBCM/Sapphire Survey 2021 p. 53, 2022 p. 52, 2023 p. 30, 2024 p. 37 (average allocations)Verified

Two consequences follow. First, the correction cut through this line. When operating discipline forced S&M down from 54% toward a projected 32% of revenue, already 37% by the 2024 actual (see Operating Expenses), companies were reducing a budget that, in part, contained the majority of their retention capacity. A cut recorded as acquisition efficiency was, in part, a cut to Customer Success that no financial statement would ever show as one. Inference

Put a number on it. A $25M-ARR company cutting S&M from 54% of revenue to the 2024 actual of 37% cuts roughly $4.25M ($13.5M down to $9.25M). At the survey’s own 57–68% CS-inside-S&M allocation range, somewhere between $2.4M and $2.9M of that cut came out of a budget line no financial statement ever labeled Customer Success. Calculated

Second, the allocation choice moves headline metrics. In the one year the survey tested it, companies booking CS in S&M reported 80% subscription gross margin against 76% for those booking it in COGS, and 41 of 75 respondents took the S&M route. The same choice contaminates CAC and magic-number benchmarks (see The CAC Trap): for a majority of companies, “S&M” is not pure acquisition spend, and by different amounts per company.

Source: KBCM/Sapphire Survey 2022, p. 52 (gross margin by CS-allocation bucket; company mix differs across buckets)Verified

The Load Curve

While its cost stayed invisible, the function’s workload was repriced. The CSM’s book of business rose from $1.7M in 2022 to $2.1M in 2023 to $2.5M in 2024, a 47% increase in two years, while median accounts covered fell from 30 to 21.5. A labeling caveat travels with the 2022 anchor: the survey titled that chart a median but labeled the bars as averages, and later editions print medians, so treat the 47% as indicative rather than statistic-pure. The direction is not in doubt. Fewer, larger accounts per CSM: the implied ARR per covered account roughly doubled. One qualifier belongs in the record. The Enterprise book was flat across the same window, $2.6M to $2.5M, so the overall jump is at least partly a composition effect, the respondent pool and coverage shifting toward larger books, rather than every CSM’s book inflating. The direction is still the same at the segment level: Enterprise accounts per CSM fell from 14 to 10.5 on a flat book, which is fewer, bigger accounts by another route.

Source: KBCM/Sapphire Survey 2023 p. 21, 2024 p. 16, 2025 p. 30 (medians, Overall). Metric labels changed between editions (2023: “CSM Accounts Covered” / “CSM Avg Book Size”; 2024+: “CSM # of Accounts” / “CSM Book of Business”); values joined as one trend with that caveat.Verified

This repricing ran concurrently with the retention decline: NDR went 106% to 102% to 101% over the same three data years (see The NDR Crisis). The pairing is correlational. Two overlapping series across three years cannot establish that rising CSM load caused falling retention, and this report does not claim it. What the pairing does establish is that per-CSM revenue responsibility rose sharply during exactly the period when the expansion those CSMs are responsible for was weakening. Inference

The Counter-Evidence

The convenient conclusion here would be that SaaS companies starved Customer Success. The data does not support it, and the one org-structure snapshot that exists points the other way.

Source: KBCM/Sapphire Survey 2025, p. 29 (sales headcount mix, n=38)Verified

The Services Gradient

The survey does contain one strong signal about what high-touch investment correlates with. In the three years it published churn across the full services-attach gradient, the heaviest-attach cohorts (professional services priced at over 50% of the contract’s own upfront value) churned at a fraction of the no-services rate: 5.8% against 16.3% in 2018, 9% against 15% in 2020, 5% against 13% in 2021. The record is not uniform: the two later editions compressed the bands, and in the 2022 data the heaviest published cohort, over 15% attach, churned 11% against 15% for the no-services cohort, a much narrower gap.

The narrowing does not stop there. The following year’s edition (KBCM-2024, 2023 data) continues this exact four-band construction and documents the relationship reversing outright above 15%: churn rises again rather than continuing to fall. See The Professional Services Optimal Zone for that finding in full; read together, the two pages describe one trend continuing to its next data point, not two that disagree. Verified

This is a correlation with known confounds; heavy services attach travels with enterprise deal sizes, longer contracts, and stickier deployments, and the top buckets are small samples. But it is the closest thing the survey offers to a measurement of what deliberate post-sale investment is associated with, and in the full-gradient years the association is large.

Source: 2018 medians (small n in top buckets): KBCM/Sapphire Survey 2019, p. 56; 2020 medians: Survey 2021, p. 28; 2021 medians: Survey 2022, p. 27; compressed 2022-data bands: Survey 2023, p. 13; reversal documented: Survey 2024, p. 12Verified

The report’s conclusion (see The Imperative for Retention) argues that retention is everyone’s concern and no one’s number. This deep dive is the empirical form of that claim. The function most directly accountable for retention appears in the industry’s benchmark system as an allocation footnote inside other departments’ budgets: never measured as a spend level, cut whenever S&M was cut, and loaded with 47% more revenue responsibility as the retention it owns declined. An industry gets the visibility it builds. For Customer Success, it built none. Inference