How to Normalise EBITDA Before You Apply a Multiple

Most valuation disputes in private M&A are not disputes about the multiple. They are disputes about what the multiple is applied to. A sector multiple is a reasonably public quantity; normalised EBITDA is a private one, assembled line by line, and it is where most of the negotiable value sits.

Why normalisation exists at all

Reported EBITDA describes how a company was run. It reflects the owner’s salary decisions, the group’s tax position, the building the founder’s holding company happens to own, and a dozen accounting policy choices made for reasons that have nothing to do with the operating business. A multiple drawn from a peer set describes something else: a professionalised cost base, an arm’s-length rent, a market-rate management team.

Applying the second to the first compares two different things. Normalisation is the work of making the denominator match the multiple, so that what you end up with is the value of the business rather than the value of a particular owner’s arrangements.

The adjustments almost every private company needs

Five categories account for most of the gap between reported and normalised EBITDA in an owner-managed business. Each one needs a specific piece of evidence, not an assertion.

• Owner and family compensation to market rate. Owner-managers pay themselves for tax reasons, not market reasons, sometimes far above a market salary, sometimes far below it while taking dividends. The adjustment is the difference between what was paid and what an external hire in the same role would cost. Evidence: a salary benchmark for the role, region and company size.
• Related-party rent and services to arm’s length. Where premises, IP or shared services come from an entity the owner also controls, the charge is a negotiating position, not a market price. Evidence: a comparable commercial lease or a transfer-pricing study.
• Genuinely non-recurring costs. A litigation settlement, a restructuring, a one-time ERP migration, the costs of the transaction itself. Evidence: the settlement agreement, the project invoice, the provision in the accounts.
• Discontinued operations and closed product lines. If a segment will not exist after completion, its losses should not depress the earnings a buyer is pricing, but only once closure is actually decided and costed.
• Personal expenses running through the P&L. Vehicles, travel, memberships. Small individually, and almost always the first thing a buyer’s accountant finds.

Adjustments run in both directions. A one-off government grant, an insurance recovery or an unusually favourable supplier settlement inflates reported EBITDA and has to come out. A schedule that only ever moves the number upward tells the reader something about the preparer rather than about the business.

The adjustments that are genuine judgement calls

Beyond the routine items sit four questions where reasonable analysts disagree. What matters is less which answer you choose than whether you apply the same answer to the company and to the peer set.

• Capitalised development costs. Under IAS 38 a company may capitalise qualifying development spend; many private companies capitalise aggressively because it flatters EBITDA, while peers expense the same work. If your comparables expense development, capitalised costs have to be pushed back through the P&L before the multiple is applied.
• Share-based compensation. It is a real economic cost, and the market treats it inconsistently. The defensible position is to state explicitly whether your multiple is calculated pre- or post-SBC and to apply that definition on both sides.
• Lease accounting. Under IFRS 16, operating lease expense moves below EBITDA, which raises EBITDA and requires the lease liability to be carried in enterprise value. Comparing a post-IFRS 16 EBITDA against a multiple built on pre-IFRS 16 earnings overstates value on both halves of the fraction.
• Grants, R&D credits and FX. Recurring, policy-driven R&D credits arguably belong in sustainable earnings; a one-off innovation grant does not. Unrealised FX movements rarely do.

A worked example

A B2B software business reports EBITDA of €2.40m for the year. The owner runs it, the office belongs to a family holding company, and a long-running dispute settled in the period.

Normalised EBITDA is €2.82m, 17.5% above the reported figure. At a sector multiple of 11×, the difference between the two denominators is the difference between an enterprise value of €26.4m and one of €31.0m, €4.6m that rests entirely on a one-page schedule.

Normalise the company the way the multiple was built

This is the step that is skipped most often, and it is the one that does the most damage. A multiple is a ratio, and both halves carry a definition. If the multiple you are applying was derived from reported earnings of listed companies, and you apply it to an enthusiastically normalised private EBITDA, you have not valued the business, you have quietly moved the goalposts.

Public comparables need normalisation too, and in practice they get less of it, because the adjustments available from published accounts are limited. That asymmetry is an argument for restraint on the private side, not for aggression.

The DealMatrix multiples are derived from institutional-grade public-market index fundamentals across roughly 150 GICS sub-industries, cleaned for outliers and gaps and then adjusted for the macro cycle, region and funding stage. That derivation is documented in the methodology, and it is documented precisely so that the earnings definition on your side can be matched to it rather than guessed at.

Where normalisation goes wrong

• The recurring one-off. A restructuring in each of the last four years is not an exceptional item; it is how the business operates.
• Double counting. Removing the losses of a discontinued line and separately removing the redundancy costs of the same closure adds back the same money twice.
• Forgetting the balance sheet. An adjustment to earnings usually has a consequence in the enterprise-value bridge, a lease liability, a provision, a deferred consideration. Adjusting one and not the other breaks the arithmetic.
• Normalising to an unachievable cost base. If the owner’s €450k salary is replaced with a €270k market salary, someone has to do the job for €270k. If the business cannot recruit at that level, the adjustment is a fiction.
• Presenting the result without the schedule. A normalised EBITDA without a line-by-line schedule and its evidence is treated, correctly, as an unsupported assertion.

Normalisation is ultimately an exercise in credibility. The schedule is read by someone whose job is to find the weak line in it, and the value of the whole document is set by that weakest line.