Market Observatory · 2026-Q3

What their own capital costs

The one estimated section. What each buyer pays for capital, set against the return it demands of the companies it buys.

What their own capital costs

This is the one section that is estimated, not measured. Everything above is read from a filed account. The cost of debt below is too — it comes from the contracts in the table above, and so does the last column. What sits between them is assumption, and every assumption is printed with it.

Serial acquirers judge deals against a hurdle of roughly 20% to 30%. That is a return on one deal, before borrowing. Its proper counterpart is not a buyer’s blended cost of capital but the cost of the business risk alone — taken here as 11.1% for every buyer. One difference between them is measurable and is set out below: how widely each buyer’s own prices scatter. It is deliberately not used to move that 11.1%. The direction would be ambiguous — a buyer paying very different prices is either buying very different things, which raises its risk, or spreading it, which lowers it — and the sample cannot carry it: only 5 of 9 buyers have 4 or more prices of their own on file, and the rest have two or fewer. An invented difference would be worse than none.

Borrowing does not close that gap; it moves the risk. A buyer who funds itself with debt makes its own equity more dangerous, and what equity costs rises to match. That is why the WACC column below barely moves across buyers whose debt weights run from 16% to 70%: what is left of the advantage is the tax deduction on interest, and no more. Where a buyer earns no taxable profit there is not even that.

Buyer	Cost of debt	Borrowed in	Debt weight	Cost of equity	WACC	Earnings may fall	Equity
Bending Spoons S.p.A.	7.9%	USD/EUR	70%	16.8%	9.2%	67%	implied
Everfield Aquiline	7.5%	EUR	52%	14.9%	11.1%	none	implied
Visma Hg	5.3%	NOK/SEK/EUR/DKK	24%	12.5%	10.5%	78%	implied
Software Circle plc	6.5%	GBP	16%	11.7%	10.7%	86%	quoted

Cost of equity is derived, not assumed. It starts from the 11.1% business-risk cost — 3.0% risk-free, a beta of 1.0 on a 5.5% premium, 2.6% for size — and rises with each buyer’s own debt-to-equity. Bending Spoons S.p.A. carries the most: 16.8% .

A cost of debt is not comparable without its currency. Each rate stands on the reference rate of the money it was borrowed in, and those differ: EURIBOR at 2.5%, NIBOR at 4.7%, SONIA at 4.0%. Visma, Bending Spoons S.p.A. borrow in more than one, so the figure shown is already a blend. What looks like one buyer paying more than another is partly the two of them borrowing in different places.

The last column needs no assumption at all. It is how far earnings can drop before they no longer cover the year’s cash interest — one minus the inverse of interest cover, both figures straight from the accounts. This is where leverage shows up honestly: Bending Spoons S.p.A. has 67% of room, Software Circle plc 86%. A cheap cost of capital and a thin margin for error are not the same thing, and a buyer can have both.

Implied equity means there is no share price: the value is this register’s own median multiple applied to the buyer’s revenue, less its net debt. A crutch, and a rough one — but one that moves with the register instead of being picked. A sponsor typically requires 20% on its equity; that is a target, not a market rate, and it is deliberately not fed into the WACC.

Two buyers cannot be estimated at all, and the reasons are themselves findings.

Hawk Infinity AS — it borrows at 11.1% — as much as the 11.1% assumed here for the business risk itself. Debt ranks ahead of equity and is usually secured, so it cannot cost more than the assets behind it. The single unlevered rate used for every buyer does not fit this one, and no honest cost of equity follows from it. Its earnings can still fall 47% before interest stops being covered.
TeamSystem — its net debt exceeds what this register’s own median multiple (3.14x revenue) would value the whole company at, so no equity value can be implied. Its earnings can still fall 67% before interest stops being covered.
Chapters Group AG — the borrowings on file cover only 44% of its net debt — the rest sits at subsidiary level and is not disclosed. Its earnings can still fall 42% before interest stops being covered.
Constellation Software — only 28% of its recorded borrowings state a rate. Its earnings can still fall 94% before interest stops being covered.

How consistently each one prices

Buyer	Prices on file	Cheapest	Median	Dearest	Scatter
Bending Spoons S.p.A.	5	1.08x	2.10x	3.26x	38%
The Access Group	5	4.02x	5.46x	7.83x	26%
Everfield	4	1.33x	1.72x	2.11x	25%
TeamSystem	4	3.14x	6.81x	13.95x	65%
Visma	4	4.22x	6.04x	6.70x	19%
Hawk Infinity AS	2	0.77x	1.74x	2.71x	too few
Topicus.com	1	2.08x	2.08x	2.08x	too few
Banyan Software UK	1	2.36x	2.36x	2.36x	too few
ClearCourse	1	2.63x	2.63x	2.63x	too few

Measured, with no assumption in it: every figure is a price this buyer actually paid, from the table at the top of this page. Scatter is the standard deviation as a share of the mean, shown only where a buyer has at least 4 prices on file — below that it is the distance between a couple of points, not a distribution. This is the table that would be needed to give each buyer its own cost of business risk, and it is also the table that shows why that cannot yet be done honestly.

The rest of the register

Guess what serial acquirers paid →What European software serial acquirers actually pay, from filed accounts. Guess the price, then see the source.

What was paid, per unit of revenue →Every disclosed acquisition price in the register, one row per observation, coloured by how well it is evidenced.

How much they deploy →How much capital each European software serial acquirer actually puts to work, and across how many acquisitions.

What they owe, and what it costs →Net debt against revenue, the interest each buyer actually pays, and every credit facility on file with its terms.

Method, limits and imprint →How every figure in this register is sourced, what the sample does not support, and the legal notices.

Figures as at 2026-09-08. Every number on this page comes from a filed account or a company’s own published report — how, and what it does not support.