PMI Alpha

The return on a roll-up is not made at acquisition. Why I deliberately give up the multiple arbitrage up front – and where the return lies instead.

The case for buy-and-build reads the same in every investment memo. You buy small units at low multiples, combine them, realise synergies and sell the resulting scale at a higher multiple. The return comes from the gap between the entry and the exit valuation.

That logic is not wrong. It is just no longer scarce.

Over the past ten years, multiple arbitrage has become the default idea. Once a strategy appears in every memo, it is priced in. The result is rising entry valuations and auctions for the same unremarkable targets.

In the same documents, post-merger integration usually appears as a side chapter. It is treated as an execution topic, not as a source of return. In my view, that is exactly where the real misjudgement lies.

Ten acquisitions, one finding

Between 2022 and 2024 I acquired ten companies and brought seven brands onto one platform: Patronas, Fixhub, DSER, MiFID Recorder, Fundsaccess, Fundhero, etops. Ten locations, five countries, three hundred employees. It ended with the sale to Pollen Street Capital.

The most expensive item during that time was never the purchase price. It was bringing the systems together.

Every company brought its own technology estate, its own data model, its own reading of the same business terms. Whatever the model listed as synergy then had to be delivered by people who were running day-to-day operations at the same time.

Underestimate that effort and you do not lose the synergy – you push it back by quarters. And with a holding period of four to five years, quarters are the real currency.

The reversal I am running now

In my current venture, consolidating the travel market, I deliberately give up the multiple arbitrage at the start.

I do not approach owners with an offer to buy. I approach them with software. The business adopts the platform before any transaction is discussed. It becomes more profitable and grows faster, and that shows up in its valuation.

At first glance this looks like a mistake. In the end I pay more than I would have had to.

The objection is fair, and anyone who cannot answer it should not take this route. The answer does not lie in hoping for goodwill but in the pricing mechanism: the valuation basis has to be agreed before the value is created. Whether through a fixed multiple on a defined baseline, an option or an earn-out structure is a matter of negotiation. Without such an agreement you really are financing your own price increase.

What I get in return is substantial. I see the business from the inside for a year before I buy it – with live data instead of prepared documents. I know whether the team is on board. And most of the integration is behind me before closing.

The return thus shifts from the purchase to the integration. Exactly where most roll-ups lose it.

What has changed technically

Until recently, this approach had a hard limit. You could only integrate companies whose technology estate matched your own. The tech stack was the first exclusion criterion on every longlist.

That limit is now falling.

The hardest part of merging systems was never the programming. It was the understanding: what does this system actually do, why was it built this way, which business rule sits behind an exception nobody can explain any more.

That knowledge rarely lives in the documentation. It lives in manuals, concepts, ticket systems, source code, comments, training material and requirement documents from fifteen years.

These sources can now be opened up by machine. From them, the business logic of a system can be reconstructed and rebuilt in a different technology – not as a migration in the classic sense, but as a new build based on the extracted domain knowledge.

With that, the question of the technology estate loses its power to exclude.

Why this is where the alpha lies

This is where the circle closes back to the buy side.

If tech-stack compatibility remains an exclusion criterion for most buyers but not for me, then I am buying in a market with fewer bidders.

Companies with ill-fitting or outdated technology are regularly marked down in processes or not looked at at all. They are hard to sell, because the obvious acquirer would be a strategic buyer with its own system – and that buyer is exactly the one who shies away from the integration.

This is not a niche phenomenon. In fragmented software markets and in every industry with grown specialist solutions, it is the norm.

Whoever can integrate these targets buys market share at prices that reflect the technical difficulty – without carrying that difficulty in full.

That is the return I mean. It does not come from the gap between two multiples, but from the fact that the set of companies I can buy differs from the set others can buy.

Where caution is needed

Two caveats belong here, and they are not formalities.

First: not every data source may be opened up just because it is technically accessible. Documentation, source code, manuals and requirement documents are uncritical. Internal communication and employee mailboxes are not. Employee data protection applies there and, in Germany, works council co-determination rights when introducing systems capable of monitoring behaviour or performance. Before closing, there is also the antitrust question of information exchange. Anyone who works carelessly here creates exactly the risk they will have to explain at exit.

Second: a reconstructed business logic is a hypothesis, not a specification. It has to be checked by someone who understands the business. The acceleration lies in opening things up, not in deciding.

What I advise investors

Anyone investing in buy-and-build strategies today should weight three questions differently than five years ago.

Not: at what multiple is it being bought. But: how large is the range of targets this buyer can actually integrate.

Not: which synergies are modelled. But: how long until the first fully combined unit.

Not: how experienced is management at buying. But: what has it integrated, and how long did it take.

Buying skill is common. Integration skill is rare. And in a market where everyone chases the same arbitrage, the rare thing is what gets paid for.

---

Johann Horch is founder and managing director of promerget GmbH, an AI-native M&A boutique for the mid-market in the DACH region, and Chief Growth Officer of Voyager AI GmbH. From 2017 to 2025 he was CEO of niiio finance group AG and acquired and integrated ten companies between 2022 and 2024.