The party writing this down would prefer the takeover to happen. That is worth stating before anything else, because what follows is a list of reasons a takeover goes badly, and a supplier with an interest in the sale is an odd author for such a list. The alternative is that the list stays private until week three, which is exactly when it costs the most.
The first thing that breaks is access, and it breaks in a specific way. What arrives at handover looks like a list of logins and is actually a list of assumptions. The advertising account turns out to sit inside a former agency's business manager, and the company holds a user seat on an asset it does not own. The domain sits with a developer who built the site four years ago and now answers slowly or not at all. The analytics property is attached to a personal address belonging to someone who left. None of this is malice. It is the residue of every arrangement the company has ever made on its digital side, and nobody kept the map.
The cost is time, and it lands in the worst possible place, which is the beginning. The build list cannot be finalised while the inventory is still uncertain, and an engagement that starts inside a fog of unknowns produces its first weeks of work in the wrong order. It is handled by treating access as something to be inventoried rather than collected. Every account is checked for ownership rather than for whether the password works, and where ownership cannot be recovered the asset is rebuilt instead of chased. Rebuilding is written into the contract as a named deliverable with a fee attached, which also means that if it does not happen, that fee comes back.
The second thing that breaks is historical data, and it breaks quietly. Tracking installed twice with overlapping events. A replatform that reset the order history. Events renamed halfway through a year by someone with good intentions. A currency change. Any one of these makes the previous period unreadable, and a period that cannot be read cannot serve as a baseline. The cost is not that reporting comes out wrong. The cost is that honest reporting goes thin precisely when a new client most wants to see something.
It is handled by refusing to smooth it over. The reported number is revenue divided by directed spend, and the published rule is that where data is missing, the attribution chain is broken, or the period is too short, the number is not shown and the reason is written in its place. An empty slot is unpleasant to look at. It is also the only version of the report still worth reading in month six.
The third thing that breaks is a person, and this is the expensive one. Somewhere inside the company there is usually someone whose job overlaps the takeover. They send the email. They run the social account. They brief the freelancer who makes the images. The takeover removes the content of that job. Very often nobody has told them, and they find out when an access request lands in their inbox.
What follows is not sabotage. It is slowness. Passwords are being looked for. The right person is on leave. The export will come on Friday. Context that lives in one head stays in that head. An engagement can lose several weeks to a sequence of individually reasonable delays, and not one of them will ever be described as a refusal.
It is handled by naming the overlap before the contract rather than after it. The scoping conversation asks who currently performs each function, what remains of that role afterwards, and whether the person holding it knows. Access requests then go to the owner rather than being negotiated with the person losing the work. Where a role continues in changed form, it is written into the same document as the deliverables, because an undefined role standing next to a defined one is a role that will be defended.
The fourth thing that breaks is an expectation, and it belongs to the owner. Building comes before running, because there is nothing to run until the missing pieces exist. The first weeks therefore look like construction: a store being rebuilt, a tracking layer being laid, a sending domain being warmed, a brand foundation being written. Measured against the steady state described on the call, that looks like nothing happening.
The cost of the mismatch is pressure, and pressure produces bad decisions early, which is the worst moment to make them. It is handled by making the build legible. The build list becomes the contract's deliverables item by item, so progress in the first weeks reads as named items completed rather than as a performance to be judged. It is also handled by saying, before any money moves, that the first weeks will look like this.
One of these four cannot be resolved in advance, and it is the person. Access can be audited, and the audit is uncomfortable but finite. Historical data can be inspected once access exists, and its defects can be listed. The owner's expectation can be set on the call, in writing, with the build list attached to it. But what a colleague believes their job to be, and what they will do when they conclude that it is being taken, is not observable from outside the company. That person is frequently not in the scoping conversation. When they are, they are answering in front of the person who employs them, which is not a condition under which anyone says the true thing. It surfaces only once the change becomes concrete to them, and the change becomes concrete only after the engagement starts. That one is discovered by starting, and the only genuine mitigation is that the owner decides in advance how it will be handled when it appears.
If someone inside the business is going to feel this takeover before it helps them, name their role on the call, because that is the failure mode that costs the most and the only one that cannot be found from outside.