The model: own a boring niche completely
The common thread is niche dominance most investors never notice. Rational owns ~50% of the global combi-oven market; Geberit is a near-monopoly behind bathroom walls; Halma, Judges and Spectris each own clusters of micro-niches in safety and measurement. Because these niches are small, specialized and unglamorous, incumbents face little big-company competition — and that buys pricing power, which shows up as the fat, durable margins that define the cluster: GEA at 16.5% EBITDA and a remarkable 36.2% ROCE, Geberit at 29.4% EBITDA, Rational at 26.4% EBIT with net cash, Nemetschek at 31.2%. The cross-cut’s first lesson: the most reliable compounding hides in markets too small and too boring to attract a fight.
Two ways to build it: the operator and the acquirer
The seven split into two distinct architectures of the same idea. The focused operators — Rational, Geberit, Nemetschek, GEA — own and deepen one domain, compounding organically through pricing, service attach and conversion of their niche. The serial acquirers — Halma, Judges, Spectris — are the opposite: decentralized roll-ups that buy niche leaders (Judges at ~5× EBIT versus the 12–14× trade buyers pay — the discount itself is the return) and run them hands-off. Halma has bought ~150 businesses since 1983. Reading the operators beside the acquirers shows two routes to the same fortress economics: build the moat, or buy a portfolio of small ones.
The recurring layer is what survives the cycle
Every one of these leans on a recurring annuity to smooth lumpy hardware. Nemetschek is the extreme — 92% recurring software revenue, the “AEC Adobe.” GEA’s ~40% service mix is what lets a maker of lumpy dairy plants hold 36% ROCE; Geberit’s ~60% renovation exposure cushions the new-build cycle; Rational’s ~90% repurchase rate on aftersales is the razor-and-blade engine. The pattern across the cluster: the quality of a compounder is largely the quality of its recurring layer, because that is the part that keeps paying when the capital-equipment or construction cycle turns down — which, for most of these, it has.
Two strains, shared by all: the scale ceiling and the price
Read together, the studies expose where the model breaks. First, the scale ceiling, sharpest for the acquirers: Judges, at just ~£280m, can no longer find deals big enough to move the needle (only £1.9m deployed in 2025), and even Halma’s pace likely slows from a larger base. Compounding by small bolt-ons gets mathematically harder as you grow. Second, the valuation: quality this visible gets fully priced — Halma at ~39× forward (richest in its cohort), Rational ~28×, Geberit ~27×, Nemetschek de-rated from ~62× but still rich. The cluster’s recurring warning is that a wonderful business at a demanding multiple can still be a poor investment — and several here (Judges −65%, Nemetschek −55%) have already shown the air-pocket when growth merely decelerates.
Where they agree — and where they split
All seven share the thesis that niche dominance plus a recurring annuity compounds durably, and all carry a cyclical end-market (labs, construction, food capex) under the quality veneer. They split on the two strains. On durability of the engine: the operators debate saturation (Rational already owns half its market; Geberit faces a weak European cycle), the acquirers debate the scale ceiling and thinning deal supply. On price: the bull reads the multiple as fair for compounding quality; the bear reads it as pricing near-flawless execution — and Spectris is the cautionary endpoint, where a public “quality compounder” trading cheaply was simply taken private by KKR at a 96% premium. The quality of these businesses isn’t the question. Whether the model keeps compounding at scale, and whether you overpay for it, is.