The Year After an Acquisition Is When Both Companies Forget Why

The deal transfers the assets and leaves the arguments behind

On the day a deal closes, a long list of things changes owner. Contracts, code, customer accounts, headcount, a roadmap, a pricing table, a support queue, a set of vendor relationships. All of it is written down somewhere, because months of diligence forced it to be written down.

The reasoning does not change owner. Why the pricing table has an odd third tier that only four customers sit on. Why enterprise support tickets route through a named person instead of a rule. Why the roadmap has a six month gap in the middle of last year that nobody has explained. Each of those has a reason, and each reason was worked out in a meeting the acquiring company was never in.

For a few months this costs nothing. The acquired team is still there, still reachable on Slack, still happy enough to answer. The bill arrives later, when somebody reaches for a reason and finds only the artifact it produced.

This is a different problem from losing one expert. A company that acquires another company inherits a whole second past, with its own vocabulary and its own settled arguments, and it inherits that past in a form it cannot search.

Three possible origins, and no way to tell them apart

Two years after a merger, the question of why something works the way it does has three plausible answers. It is how the acquirer always did it. It is how the acquired company always did it, and it survived. Or somebody decided during integration that this is how it would be from now on.

The right response differs in every case. A practice carried over from the acquirer because nobody thought to question it deserves questioning. A practice the acquired company held onto for a specific reason deserves understanding before anyone removes it. A call made during integration under time pressure with half the facts available deserves a deliberate second look.

From inside the combined company, all three look the same. They present as how things are done here, with the same air of permanence. The safe move becomes to leave everything alone, which is how a merged organization ends up carrying a set of practices that nobody in the building can defend or explain.

None of this shows up on a status dashboard. It shows up as a slow tax on every proposal, because anyone arguing for a change has to first do archaeology on why the current thing exists.

Integration decisions are the least recorded decisions a company ever makes

Melissa Graebner, Koen Heimeriks, Quy Nguyen Huy and Eero Vaara reviewed the research on post-merger integration for the Academy of Management Annals and counted more than 300 articles published on the subject since 1985. Their finding about decision-making is the one worth reading twice. The literature, they write, offers little insight into how integration decisions are actually made, and with few exceptions researchers have not examined the process of integration decision-making in real time or in a fine-grained manner.

That is a statement about what academics can see, and it lines up exactly with what a combined company feels a year later. These decisions get made quickly, by a group assembled for the purpose, during a stretch of months when everyone involved is doing this on top of their real job. Then the group dissolves back into the organization and no standing team owns the record.

The same review reports Barkema and Schijven's argument that initial integration decisions will be suboptimal because of information asymmetries and bounded rationality. What follows from that is the part with teeth: they found that each additional acquisition raises the hazard of a later restructuring, so the early calls do not stay contained inside the integration period.

If the first round of decisions was made with information missing, then revisiting them later requires knowing which information was missing at the time. That is a record problem before it is a strategy problem.

A default looks exactly like a decision once it is a year old

The same review quotes Haspeslagh and Jemison on how firms automatically impose their administrative systems and practices on the acquired firm without considering whether those systems are right in the new setting, and Mirvis and Marks describing a sense of superiority in which the acquiring company's leaders assume their own procedures, policies and systems are better than the ones they bought.

In practice nobody announces this. The acquirer's expense tool wins. Its planning cadence wins. Its incident severity levels win. No comparison is run, because running one would take a week and there are ninety other open items, several of them with a legal deadline.

Twelve months on, the record shows the acquirer's process operating across both companies. It reads identically whether the choice was argued or simply assumed. The person who now wants to change it is arguing against something that has acquired the appearance of a settled decision.

The fix at the time is one line in the notes: this is the practice we adopted, this is the one we set aside, and here is whether we actually compared them. Reconstructing that a year later is expensive and usually fails, because the people who would know have started answering with what they think they remember.

The people who can explain the other company leave first

There is a measurement of how fast this window closes. Using employee-employer matched data from the US Census, a study of roughly 4,000 high-tech startup acquisitions in the United States between 1990 and 2011 tracked around 300,000 non-founding employees at the acquired firms against roughly two million people hired at the same acquirers in the same year.

In the first year after the acquisition, 34 percent of acquired workers left the firm, compared with 12 percent of the matched regular hires. Across a three-year window, acquired workers were about 15 percent more likely to leave. The gap is at its widest in year one and narrows steadily after that.

One detail in the results matters more than the headline number. The effect was stronger among high-earning individuals, which the paper describes as implying a loss of core organizational knowledge from the acquisition. The people most likely to be carrying the reasoning are the people most likely to walk.

Now set that against how long the work takes. The Graebner review cites Birkinshaw and colleagues finding an initial three to five year period in which human integration is the priority, and Allatta and Singh finding that communication between acquirer and target staff rises gradually, peaks around two years after close, then falls away. The period when you can still ask somebody ends well before the period when you still need to.

Two archives, one search box

Both companies arrive with a pile of notes, docs and recordings. The obvious move is to put everything into one index so the combined company can search its combined history. The obvious move creates a specific and quiet failure.

The problem is vocabulary. The platform means one system at one company and a different system at the other. A priority one incident is a middle-of-the-night page at one and a fix-it-this-week label at the other. Enterprise is a price band in one archive and a team name in the other. Two customers have similar names and one of them churned. A search across the merged pile returns an answer that is completely accurate about the wrong company.

That failure is quieter than an empty result. A blank search sends somebody off to ask a person. A confident answer with the wrong provenance ends the enquiry, and the mistake surfaces a month later in a customer call.

So keep provenance attached to everything: which company a note came from, what date, who was in the room. Searching across both is useful. Flattening them into an undifferentiated pool is not. For the first year, a cross-company answer is worth treating as a lead to check rather than a fact to act on.

The integration meetings are the ones worth capturing properly

If a combined company can only get one class of meeting right, it should be the integration meetings themselves. They are short-lived, high-consequence, and attended by people who are about to disperse. Nothing else in the calendar has that combination.

What to keep from each one: what was chosen, which company's practice it came from, what the alternative was, the constraint that forced the timing, who owns it now, and the date it is worth looking at again. That last field does most of the work, because it converts a rushed call into something with a scheduled second look rather than a permanent one.

The other sessions worth capturing carefully are the ones where the acquired team walks through its own systems and accounts. Those usually happen once, early, in front of two or three people, and they are the nearest thing to a real transfer of reasoning that the whole process contains.

Then there are the boring recurring ones. A weekly integration standup captured across six months is a better account of how the two companies actually came together than any summary deck produced at the end. The deck records the conclusion. The standups record the order things were tried in, and which assumptions turned out to be wrong.

How Driffle fits, and what it does not solve

Driffle transcribes audio directly from the computer, so no bot joins the call and no notification goes out to the other side. In the months after a deal that matters more than usual, because a good share of those meetings involve people who have just been told their employer changed and are watching closely for anything that looks like surveillance.

Nothing auto-joins, auto-records or runs in the background. You start it and it works, you close it and it stops. Audio is transcribed in real time and discarded, and what is kept is the text and the notes. Notes stay visible only to their author until they are explicitly shared, and workspace admins see what has been shared into shared folders rather than private notes.

Afterwards you can ask questions across past meetings, pull out action items and owners, and send notes into Slack, email, Notion or a CRM. For an integration lead running twenty threads at once, the useful part is being able to ask what was decided about a specific system in March and get the answer without finding the person.

The limits are worth being straight about. None of this reaches back before the deal, so the acquired company's last four years of reasoning are only available to the extent that team captured them or is still around to explain them. Capturing an integration decision does nothing to improve the decision. The value arrives later, when somebody can find out what it was for.

A test you can run this week

Write down the last three times somebody in the combined company asked why something works the way it does. For each one, note two things: whether the answer came from a record or from a person, and whether that person came from the acquiring side or the acquired side.

If the answers came mostly from people, and mostly from one side, the company is running on borrowed memory. That is survivable in month three and expensive in month eighteen, and the repayment schedule is set by other people's job offers rather than by anything on the integration plan.

The useful thing about the test is that it takes ten minutes and produces a number you can watch. Run it again a quarter later. If more answers are coming from records and fewer from the same two people, the integration is building memory rather than spending it.