Comparables are the method people think they are using when they mention a number of “times”. On a UK high street, in a partner meeting, or in a short note to a bank, EV/EBITDA is the dialect. It is useful. It is also how a listed group’s rating gets photocopied onto a company that could not raise equity on a Thursday afternoon.
Enterprise multiples and equity multiples
An enterprise multiple, such as EV/EBITDA or EV/Sales, is a price for the operations before the cheque is split between lenders and shareholders. An equity multiple, such as price/earnings, is a price for the shares after interest. Mixing them is a category error. If the numerator is enterprise value, the earnings measure must be before interest. If the numerator is equity value, the earnings measure is after interest, and the capital structure is already inside both sides.
For Porth Engineering Ltd, the fictional fabricator from the cash flow essay, the clean starting point is EV/EBITDA. Sales multiples are a cross-check when EBITDA is temporarily odd. Price/earnings is a poor primary tool for a small private company whose owner’s salary is a policy choice rather than a market wage.
What “comparable” has to mean
A peer is not a company in the same loose industry label. For a regional precision fabricator the relevant likeness is activity, margin shape, customer concentration, capital intensity, and the fact of being small. A global aerospace supplier with 18 per cent EBITDA margins and a 15-year defence order book is not a peer just because both businesses cut metal.
Listed peers are observable, which is why they dominate lazy sets. They are also liquid, diversified, and professionally governed. Those differences are not a footnote. They are the reason a private-company discount exists, and the reason a single percentage rarely deserves the confidence it is given.
A classroom peer set
The three names below are invented. They are not quotes, not a study, and not a claim about UK manufacturing multiples in any year. They exist so the arithmetic has somewhere to stand.
| Illustrative peer | EV/EBITDA | EV/Sales | Why it is here |
|---|---|---|---|
| Harbour Fabrication | 8.4x | 1.15x | Larger, listed, broader customers |
| Severn Components | 7.1x | 0.92x | Closer margin, still bigger |
| Tamar Precision | 6.6x | 0.88x | Regional, thinner disclosure |
| Median | 7.1x | 0.92x | The middle, not the average of a story |
The median is used rather than the mean so one optimistic name cannot drag the set. With three names the median is simply the middle figure. A real file with eight peers would still show the range, not only the middle, and would say which peers were rejected and why.
Normalised EBITDA, or the multiple is theatre
Porth’s reported EBITDA in the base year is 720,000, and for this essay that figure is already normalised: market owner salary, no one-off grants, no related-party rent games. If it were not, the work would happen before the multiple, not after. A multiple applied to a flattered EBITDA is two errors in a stack.
The sales cross-check is worth doing in the same breath. Median EV/Sales of 0.92 times revenue of 4,800,000 is an enterprise value of 4,416,000. Keep it in your pocket. If the EBITDA route produces a wildly different enterprise value, the margin you are capitalising is the thing to interrogate.
The 20 per cent confession
Apply a 20 per cent discount to the median EV/EBITDA to stand in for illiquidity, key-person dependence, customer concentration and weaker reporting. 7.1 times 0.80 is 5.68. That is the private-company multiple used here.
It is a confession because a round percentage admits you cannot observe the discount directly. Sometimes the evidence supports more than 20. A company whose largest customer is 40 per cent of revenue, and whose founder still prices every job, is not suffering from a mild liquidity cold. Sometimes the evidence supports less, for example a contracted, transferable order book and a second-tier management team that already runs the week. Write the reasons beside the percentage. If you cannot, you have decorated a guess.
The bridge, in pounds
5.68 times EBITDA of 720,000 is an enterprise value of 4,089,600. Net debt, on the same basis as the cash flow essay, is 620,000. Equity value is 4,089,600 minus 620,000, which is 3,469,600.
Set that beside the discounted cash flow equity value of 3,946,337. The gap is about 477,000, roughly 14 per cent. That is a conversation, not a failure. The cash flow model is willing to pay for a fade of growth that the multiple, taken from peers, may already capitalise differently. Averaging 3,469,600 and 3,946,337 to a single “answer” of 3.7 million throws away the only interesting information, which is why the two methods disagree.
Failure modes that show up in UK files
- Last-twelve-months EBITDA that still contains a one-off insurance receipt, a Covid-era grant, or a year in which maintenance was silently skipped.
- A listed mega-cap multiple applied to a company that could not bear listed-company costs.
- An equity multiple used on pre-interest earnings, which ignores the loan and flatters the shares.
- A discount “for being private” stacked on top of peers that were already private, so the discount is taken twice.
- Ignoring surplus property. If a workshop sits on land the trade does not need, that value belongs in the bridge or in an asset cross-check, not inside the multiple as if it were earnings.
Comparables are the right primary language when there is a real set of similar businesses and the decision is a price in the market’s current mood. They are a weak primary language when the company is genuinely idiosyncratic, loss-making for a reason you understand, or about to be wound down. In a sale process, the multiple that matters is often not a trading multiple at all. It is what buyers have paid for control. That is the next essay.