Buy first, integrate never
Roll-ups are usually described as a cycle. Buy a national hoster, move its customers onto the group platform, take the cost out, then buy the next one. The migration is where the synergy lives, so a buyer with a queue of half-digested acquisitions is supposed to slow down until the queue clears.
Public infrastructure signals show where each acquired brand's customers are actually served from, so the queue is measurable from outside. It is real. It is also not acting like a brake.
Integration is real — for two brands in three
That matters before anything else, because the interesting number is a residue, not a failure rate. The question is not whether these groups integrate. It is what happens to the part that does not.
The residue does not clear
The other 30 brands fall into two groups, and only one of them is a story. 7 were bought inside the last two years and carry 152,608 live domains between them: a brand bought last quarter is mid-integration, not un-integrated, so they are counted as neither.
That leaves 23 brands whose customers are still served from the infrastructure they were bought with, two or more years after the deal. They are not a rounding error: together they carry 1,340,436 live domains. 21 of them have moved nothing at all; the other 2 moved their mail onto the group's platform and left the hosting where it was, and the younger of those two deals is 69 months old.
Those 21 have been sitting there for a median of 49 months — just over four years. The oldest was acquired 254 months ago, a little over two decades.
The obvious objection is that these are migrations in progress, caught mid-flight. They are not. Alongside the headline reading of each brand's infrastructure there is a second one: the share of its individual domains already served from the group's platform, which is what a migration in progress looks like from outside. For 20 of the 21 that have moved nothing, that share is under 5%. The work has not started.
And it does not gate the next deal
Seven of the nine groups carry at least one of these brands. All seven announced further acquisitions afterwards — 78 validated deals between them, counted only from the date of the oldest brand each owner still holds entirely on its own infrastructure. Not one of the seven paused while its queue stood still. The remaining two groups carry no such brand at all: every acquisition we can read for them is served from a group platform.
It would be tidier if the owners were simply dormant, but the second instrument says otherwise. Over the last 90 days, between 0.45% and 21.23% of these same groups' books moved between platforms the owner already controls — the ordinary churn of reorganising a portfolio. The books are being worked. The acquired backbones are the part that stays where it is.
What the record says about integration
An integration rate is one number; the record behind it says four things.
- Which acquisitions were integrated says more than the integration rate. Two in three is a respectable number and it hides the shape: the residue is concentrated, old, and made of businesses that were bought with their own infrastructure and still run it.
- The halfway cases sit on the un-integrated side. A brand whose mail has moved and whose hosting has not is not an integrated brand; it is two platforms with one of them shared. A reported integration percentage depends on what a partial move counts as.
- "Planned migration" is a dated commitment. A brand that has moved under 5% of its customers four years after the deal has a plan, not a project.
- A parallel estate has a running cost. A parallel estate carries two renewal cycles, two support surfaces and two sets of infrastructure contracts. Where a group has carried that for years without visible urgency, the parallel estate is the steady state and the migration is the exception.
None of which says the owners are wrong. Keeping a strong local brand on its own infrastructure is a defensible strategy - customers rarely leave for it, and a migration is the moment they might. It is a strategy all the same, and the record shows it as one rather than as a synergy that arrives on schedule.
How the population is defined
Three conditions decide who is in this brief, and each of them excludes somebody. A group counts as European when more than half of its live customer book sits on a European country domain — measured, not inferred from where the company is registered, because the record holds groups registered in Europe whose customers are overwhelmingly not. Seven groups fall out on that test. A group needs a shared platform of its own: without one, a brand running its own infrastructure has nothing to be integrated onto, which is a different story, and it removes one more. And a brand is only counted where its infrastructure is readable — brands served from behind a large cloud cannot be attributed, so we exclude them rather than count an unreadable brand as un-integrated; two further groups have no readable acquisition left after that.
Reproduce this — or ask it yourself
Every figure here is queryable through the HostingBrain connector. In Claude or any MCP-compatible assistant, this is the whole brief in one prompt:
“Using HostingBrain, show me how deeply each European hosting group has integrated its acquisitions: which acquired brands still run on their own infrastructure, how long ago they were bought, and how much of each brand's book has actually moved. Then show me how much of those same groups' books moved internally in the last 90 days.”
Resolves to integration_depth and group_movement (both Pro).
definitions(term='integration_depth') explains the stage vocabulary and the weights,
on the free tier.
Reproduce this analysis: the
integration_depth tool returns the per-brand stage, acquisition date and migration
share behind every count above — Pro tier, with the free tier covering market structure,
concentration and the definitions.
Underwriting a consolidator? HostingBrain answers questions like this — with the date, denominator and caveats attached — inside Claude and any MCP-compatible assistant.