A top-quartile screen makes ICONIQ's 115% growth median a one-in-eight number

Filter any pool to its top quartile and the median of what remains sits at the 87.5th percentile of the pool you started with, while the upper quartile of what remains sits at the 93.75th ^[17]. ICONIQ's admission test is top-quartile revenue growth over the past three years, plus AI-native or AI-driven ^[4]. So the 115% median at $100M+ ^[5] describes roughly the best company in eight in the pool it came from. The 165% top-quartile figure describes roughly the best in sixteen ^[18]. The filter window and the reported growth window are not the same, so both are approximations.

ICONIQ is explicit about the design. Some of today's leading companies sit well above aggregate medians and even top-quartile figures, the firm says, so identifying top performers means benchmarking against those companies ^[11].

The pool blends top public software companies with ICONIQ's own private venture and growth portfolio, on quarterly data from 2024 through Q2 2026 where available ^[3]. SaaStr says no public software company grows 900% below $10M in revenue. That leaves the public comps able to influence the $100M+ column and perhaps the $25M-$100M one, and puts the small bands inside ICONIQ's own portfolio ^[9]. SaaStr singles out the 2600% top-quartile figure below $10M as the cell most likely to rest on a handful of companies ^[8].

Margin was never part of the screen, so the margin column reports something the selection did not put there ^[4]. The curve recovers 25 points into the $25M-$100M band and gives back five above $100M ^[19]. ICONIQ's reason is that compute and infrastructure costs have made gross margin a metric to monitor actively, and that the benchmark for a healthy margin is still moving because greater usage drives both more customer value and higher cost ^[14].

SaaStr's headline pairs the growth and margin figures with $655,000 of revenue per employee ^[15]. A company at $100M of revenue hitting that number runs about 153 people ^[20].

The caution SaaStr attaches is that the multiples being paid right now are being paid against this table ^[12]. I'd call that the defensible use of it, since a founder can work out which column a buyer is marking them against. As a plan it runs backwards. The companies in the table are there for having already grown, and SaaStr notes the index cannot tell you anything about failure rates, survival, or what a typical AI company looks like ^[10]. Disclose an n per band, with cells wide enough to include companies from outside the portfolio, and the growth medians would describe a population ^[8][9].