Operating Expenses

Faced with a softening revenue engine, the market did not repair it. It cut cost. Between 2022 and its 2026 estimate, operating expense as a share of revenue fell 43 points. The reduction lifted EBITDA by 48 points and moved no retention metric at all. This is the pivot on which the report’s second half turns: the response to the problem became part of the problem, because the savings came out of the infrastructure that produces retention.

The 43-Point Cut

Total operating expense fell from 118% of revenue in 2022 to a projected 75% by 2026, with 88% recorded in 2024. A 43-point swing in operating discipline is real and, in isolation, impressive. The important qualifier is where the improvement came from. It was not driven by growth outrunning a fixed cost base; growth slowed over the same window. It was driven by cutting the base itself, and the composition of those cuts is what determines whether the efficiency is durable or borrowed.

Source: KBCM/Sapphire Survey 2022–2025Verified

Where the Cut Fell

The reduction landed hardest on the functions that build future retention.

• Sales and marketing fell 22 points (54% to 32% of revenue), directly reducing acquisition capacity.
• Research and development fell 13 points (39% to 26%), the most consequential cut of the three, because product velocity is how a company defends against commoditization, and commoditization is exactly the pressure the AI section documents.
• General and administrative fell 9 points (26% to 17%).

Trimming G&A is ordinary discipline. Trimming R&D and the customer-facing functions to protect a quarter’s margin is a different decision: it defends present profit with next year’s differentiation and retention. Inference

EBITDA and the Rule of 40

The clearest evidence that the gain is arithmetic rather than structural is the Rule of 40. Median EBITDA swung from −47% in 2022 to roughly breakeven by 2024, a dramatic improvement on paper. Yet the share of companies actually clearing the Rule-of-40 bar fell, from 11% to 5%. The median score improved because cost-cutting lifted margins faster than growth declined, but fewer individual companies are healthy by the standard the rule is meant to measure. A median can rise while the population beneath it thins. That is what happened here.

Derived: Rule-of-40 achievement rate by yearCalculated

Flat Retention Despite the Cut

Through the entire 43-point reduction, the retention numbers did not respond. Gross churn stayed at 14–15%, the Net Magic Number held at 0.50, and net retention continued to soften, from 109% to 101%. The cost cuts bought a genuine improvement in reported profitability and bought no improvement in the underlying business. Because the reductions fell on R&D, sales and marketing, and the services that seat the product, they most plausibly made the retention trajectory worse, not better. That is the operational reaction the report’s thesis describes: a response that improved the metric reporting soonest while eroding the foundation that reports last. Inference

An apparent contradiction has to be resolved here, because a careful reader will find it. The report argues both that six years of added CS headcount and tooling failed to move the churn floor (The Half-Measures Taxonomy) and that these cuts damaged retention capacity. If added spend did nothing, why would removed spend hurt? The resolution is position, not level. The additions went downstream: more coverage against an unchanged intake of deals, applied after the determinants of churn were fixed at the close. The cuts fell upstream: R&D sets what the product can reliably deliver, and the S&M line contained the onboarding, enablement, and services capacity that determine whether a promise survives contact with deployment. Spend added after the determinants are set cannot buy the floor down; spend removed from the determinants themselves can push it up. The two findings describe one gradient read from opposite ends, and both run on the report’s 12-to-24-month lag: the deepest cuts continue into the forecast years, so their full retention effect sits beyond the last settled edition. That claim is falsifiable on schedule. If the churn floor is still 14–15% two editions after the cuts complete, the upstream-damage reading weakens and the flat-marginal-spend reading strengthens; the report accepts that test. Inference

Source: KBCM/Sapphire Survey 2022–2024Verified

When Spend Stopped Buying Growth

The natural objection to the cost-cut critique is that companies should have kept spending. The survey’s own data does not support it. In the 2018 data, sales and marketing intensity predicted growth about as cleanly as anything in the dataset: median organic growth climbed from 19–21% in the lowest spend bands to 100% in the 100–120% band, a spread of roughly 81 points from bottom to top. By the 2022 data the spread had compressed to about 25 points and the relationship had lost its shape entirely. Companies spending under 25% of revenue on S&M grew 39%, exactly matching the band spending 50–100%. The 2025 survey then quantified what remained, printing an R² of 0.315 on its spend-versus-growth scatter. In plain terms, a company’s S&M spend level now explains less than a third of its growth outcome. The cuts did not repair the engine, but the decay ran in the other direction too: by the time the cuts arrived, more spend had stopped being a reliable way to buy growth.

Source: KBCM/Sapphire Survey 2019 p21 (2018 data), 2023 p22 (2022 data), 2025 p33 (R², 2024 data)Verified

The same data locates where the remaining excess spend sits, and it is not in go-to-market. In 2024, companies growing under 10% carried a heavier cost structure than companies growing 10–20%: 85% of revenue in total operating expense against 79%. The slow growers spent slightly less on S&M (35% versus 37%) and more on R&D (29% versus 25%) and G&A (21% versus 18%). Stalled companies are not underspending on growth; they are carrying overhead. G&A is also the line the correction touched least: from the 2022 peak through 2024, S&M gave back 17 of its 54 points while G&A gave back 4 of its 26. The efficiency era cut the engine harder than the administration around it, a pattern the GTM Machine deep dive traces through the acquisition funnel itself. Inference

The comparison carries stated limits. The survey coarsened its spend bands from seven to four between the 2019 and 2023 editions and drew the two years from different respondent panels, so the 81-to-25-point compression is directional rather than point-precise. The R² figure comes from a 55-company panel. And the relationship reads in both directions: high spend bands contain companies spending into growth and companies spending against decline, so none of this establishes that any given spend level causes any given growth rate.