Every week we pull the SIC codes attached to the 650 UK companies Archive Partners tracks, and every week we default to the same question: which code shows up most often? By that count, the answer is unchanged from prior weeks — “other business support service activities not elsewhere classified” (SIC 82990) leads with 30 companies, followed by “activities of other holding companies” (64209) with 28. Between them, our tracked population spans 301 distinct SIC codes, a reasonable spread for a cohort built around mid-market UK firms rather than a single sector.
This week we ran the same data through a different filter: instead of counting how many companies carry each code, we summed the turnover behind it. The leaderboard barely resembles the one built on headcount. Holding companies (64209) still lead, but now by a wider margin — £1.62 billion combined across those 28 firms, an average of £57.9m each, well clear of anything else in the dataset. The real surprise is second place. SIC 35140, “trade of electricity,” posts £926.1m in combined turnover despite representing exactly one company in our entire tracked cohort: Pozitive Energy Ltd, a Greater London-registered supplier with £926.14m in turnover and no second SIC code to dilute the classification.
Put another way: one company, filing under one SIC code, sits ahead of 82990 — the single most common code in the dataset, carried by 30 separate firms — on the turnover leaderboard. Those 30 businesses combined turn over £679.5m. Pozitive Energy alone turns over £926.1m. A frequency table would tell you 82990 is nearly the shape of “business support services” in the UK mid-market; a turnover table tells you a single energy trader outweighs the entire category.
Neither table is wrong, but each answers a different question, and conflating them is where the mistake creeps in. Frequency answers “how common is this kind of business?” Turnover answers “how much of the tracked economy does this code represent?” A procurement team scoping supplier risk by SIC code, or a credit analyst sizing sector exposure, needs to know which question they’re actually asking. Filter by code frequency and Pozitive Energy is one row among 650, indistinguishable in weight from any other single-code filer. Filter by turnover and it single-handedly reshapes the top of the table — a reminder that a SIC-code count, on its own, tells you how many businesses use a label, not how much economic weight sits behind it.
The construction codes that have anchored this series in prior weeks make the same point from the other direction. 41100 (development of building projects), 41201/41202 (residential and non-residential construction) and 43999 (other specialised construction) collectively span 77 tagged instances across the cohort and a combined £1.79 billion in turnover — genuinely broad-based, both in company count and in economic weight, unlike 35140’s single-firm spike. That’s the difference between a sector that’s actually large and a sector that merely looks large because one outlier happens to carry its SIC code. Distinguishing the two requires pulling both numbers side by side; either one alone will mislead you in a different direction.
For a data set this size, the lesson generalises: any code held by a handful of companies is one large filing away from a misleading rank. With 301 distinct codes spread across 650 firms, the median code is carried by only two or three companies — plenty of room for a single outsized turnover figure to distort a sector’s apparent importance if you’re only looking at pounds, or to hide real concentration if you’re only counting logos. The fix isn’t complicated — cross-check frequency against turnover before drawing a sector conclusion from either alone — but it’s a step that’s easy to skip when a leaderboard already looks intuitive. This week’s data is a case study in why that check is worth the extra minute: the code with the second-highest economic weight in our entire tracked population belongs to a single company, and no amount of counting rows would have surfaced that on its own.