A single blended gross margin figure answers a narrower question than it looks like it does. Two businesses can report the same overall margin while one earns it evenly across the book and the other earns all of it on a handful of jobs, carrying the rest at a loss the blended number hides completely.
Key takeaways
Nothing about a blended margin figure is false. It is simply answering a coarser question than the one a buyer actually needs answered, which is where, specifically, the business makes its money. deavo.ai’s guide to reading financial statements before a purchase flags this directly among its red flags: “margins that move significantly from year to year without an obvious explanation”. A shift like that is often invisible at the blended level and only becomes legible once the same revenue is split by product or service line, because a mix shift — more of the low-margin line, less of the high-margin one — produces exactly the same blended-margin movement as a genuine cost problem, and the two call for completely different responses.
The first is related-party pricing. Where a management fee, a supply arrangement or an intercompany charge runs between the target and a business the owner also controls, that charge is frequently allocated to — or away from — a specific line rather than spread evenly, which can make one line look artificially strong or artificially weak. This is the same mechanism covered in related-party transactions inside the accounts, applied here at the line level rather than the whole-business level.
The second is inconsistent cost allocation, which does not require anyone to be doing anything improper. deavo.ai’s own note on this is worth taking at face value: “two accountants reviewing the same books can reasonably normalize a few borderline items differently”. Labour that serves more than one line, shared equipment time, and overhead that is not tracked by job all have to be allocated using some method, and a small, well-intentioned change in that method from one year to the next can move a line’s reported margin by several points without any change in what actually happened on the ground.
The tax return is the wrong document to start from — it reports one number for the whole corporation and was never built to be un-blended. The internal job-costing or line-level accounting records are the only place this analysis can start, and they need to be tested for two separate things: whether revenue and direct cost were captured consistently by line in the first place, and whether the allocation method for shared costs was applied the same way across every period under review. Where the target genuinely cannot produce line-level cost data, that absence is itself a finding worth documenting — it means the margin figure being relied on for pricing has not actually been tested at the level the deal is being priced on.
Two businesses with the same reported revenue can have completely different risk profiles behind that number, which is exactly why segmentation matters beyond the margin question alone. As deavo.ai puts it: “two businesses can report similar revenue while one earns it from a handful of large, price-sensitive contracts and the other earns it from hundreds of smaller, recurring customers.” Line-level margin testing and customer-concentration testing are, in practice, the same exercise looked at from two angles.
Margin testing is more demanding where revenue is recognized on partly finished work rather than completed sales. A contracting business’s job-level margin depends directly on how work in progress is valued and how holdback amounts are treated, which is covered separately in work in progress on a contracting business — read the two together where the target does project-based work, since a WIP valuation error and a margin-testing error tend to be the same underlying issue.
Segment-level margin testing and the normalization work covered in related-party transactions inside the accounts draw on the same underlying records and are easiest to do as one pass rather than two. An add-back that adjusts owner compensation or a related-party fee changes the earnings base a multiple gets applied to; deavo.ai’s explainer on SDE versus EBITDA is explicit that this earnings base, not the margin figure itself, is what a multiple is actually priced against — “a multiple applied to SDE is not directly comparable to a multiple applied to EBITDA for the same business, since the earnings base itself is calculated differently.” Getting the line-level margin right and getting the add-backs right are, in practice, one exercise: an unsupported add-back at the whole-business level almost always turns out to sit disproportionately inside one line once you look, which is a second reason to test margin by line rather than settle for the blended figure a seller’s summary presents.
Suppose a target reports a blended gross margin of 34% across three service lines, and segmentation shows line A at 44%, line B at 38%, and line C at 9% — illustrative figures only, set here to show the process. The blended number alone gives no reason to look further. Once split, line C is either a genuine problem (underpriced relative to its cost to deliver, or carrying an allocation of overhead the other two lines are not) or a strategic loss-leader the owner can explain by reference to a specific customer relationship it protects. Either answer changes how the deal should be priced; neither is visible in the blended 34%.
Treat that as a finding in itself, not a dead end. It means the margin figure used to price the deal has not been tested at the level the deal depends on, and it is a reasonable basis for asking the seller’s accountant to help reconstruct at least an approximate split for the periods under review.
Far enough to see a trend, not just a snapshot — two to three years is the usual minimum, matched to whatever period the rest of financial diligence is covering.
No. A deliberate mix shift, a strategic pricing decision, or a one-off cost in a single period can all produce the same pattern as a genuine deterioration. The point of testing by line is to be able to tell the difference, not to treat every decline as disqualifying.
A short call is enough to scope a line-level margin review before you rely on a single blended number.
Nobody publishes Canadian transaction data, so every valuation in this country quotes an American benchmark. We are building the Canadian one — multiples, asking-to-sale spreads and days on market, by sector and by city. Leave an email and you will see it first.
No pitch, no listings. One email when the first report lands.
