Gross margin is now a benchmark metric, and a great magic number can mean you under-invested in marketing

ICONIQ has replaced its Enterprise Five Scorecard with a new benchmark set it published on 17 September, the Pacesetter Index. It adds gross margin and gross retention, and it reinterprets a metric it kept in a way that should interest anyone whose budget gets defended in a board meeting: an excellent net magic number is no longer automatically a compliment. At very high growth rates, the firm writes, "an unusually high number can signal high GTM efficiency when it actually reflects more room to invest in GTM." That is a venture investor telling founders the number they have spent a decade using to justify cutting go-to-market spend may be evidence they spent too little of it.

What the index is, and what it is not

The cohort, ICONIQ says, is AI-forward companies that ranked top-quartile or better on revenue growth for their size band over three years. Burn multiple, gross retention and gross margin join revenue growth, net retention and revenue per FTE. Each addition has a reason. Gross retention is there because contracts are shorter and a proof of concept has become the normal way in, so revenue can walk out of a door net retention never shows. Gross margin is there because inference and infrastructure now sit inside cost of revenue, and the firm concedes its companies often run low and that the healthy level is still moving.

The composition matters before you quote any of it. SaaStr's Jason Lemkin, who published the first detailed read of the figures, describes the pool as ICONIQ's own venture and growth portfolio alongside the leading public software names, on quarterly data from 2024 into the second quarter of 2026. The benchmark is therefore drawn partly from companies the benchmarker owns a piece of, and it is top-quartile by construction. ICONIQ's page carries boilerplate calling the views proprietary research that "may be subjective." None of that makes the numbers wrong. It does mean this describes outliers, not an industry average, and a planning slide that calls it the standard is misrepresenting it.

The margin number quietly re-prices your entire funnel

Lemkin's write-up puts median gross margin at 55% below $10M ARR, 60% between $10M and $25M, recovering to 80% in the $25M-$100M band. Those medians are in his article; ICONIQ's public page does not publish them.

That is not a finance detail. CAC payback is gross-margin weighted, and almost every payback target in B2B software was set when 80% was the floor of respectability. Hold acquisition cost and contract value constant, drop the margin from 80% to 55%, and the same customer takes roughly 45% longer to pay for themselves. A board that keeps its twelve-month payback rule while margin falls to 55% has, without saying so, asked marketing to acquire customers about 31% cheaper than last year.

Most teams will never be told that is what happened. They will be told the payback target is unchanged and the pipeline number went up. Do the arithmetic yourself before planning season: it is the difference between a channel that stopped working and a channel that was quietly re-scored.

The efficiency metric you were judged on has a denominator, and someone else changed it.

Retention stops being a customer success line item

The gross retention addition has the longest tail. Lemkin reports median gross dollar retention of 90% at $100M+, so a tenth of the revenue base leaves each year even among the strongest performers, and net retention below $10M ARR at only 105%. That second figure punctures a familiar story. Consumption pricing is supposed to expand by itself as usage climbs; at the earliest stage, on this data, it barely does.

The consequence for a marketer is structural. Once gross retention is reported separately, an expansion narrative no longer covers a leak underneath it. A company at 120% net retention and 88% gross retention is growing on a shrinking set of accounts, and that is now visible to anyone reading the same scorecard. Retention messaging, onboarding content and the unglamorous work of making a product hard to leave move out of a customer success budget line and into the numbers your company is graded on.

What to do with it

• Recalculate CAC payback at your actual current gross margin, not the one in the model from two years ago, and bring both numbers to planning.
• If your net magic number looks exceptional, say so out loud and cite ICONIQ's reading of it. A high number is now defensible grounds for asking to spend more, not less.
• Ask what your gross dollar retention is. If nobody in marketing knows, that is the finding.
• Label it correctly wherever you cite it: top-quartile AI-forward companies, partly the publisher's own portfolio, not the market.

One more figure is going to travel further than it should. Lemkin puts median revenue per employee at $655,000 at $100M+ ARR, against roughly $200,000 to $250,000 historically, and he is candid that he runs SaaStr with three people and a fleet of agents. Expect it to be pointed at your headcount request within the quarter. The rebuttal is in his own piece: revenue per employee lags decisions made a year and a half earlier about how the team works, and no hiring freeze this quarter produces it.