Hosting's second act: three business models are emerging from the same customer base

Finance houses and investors are continually asking about web hosting — and the question underneath the diligence calls is not "who is biggest". It is: what kind of business am I underwriting? High-margin subscription infrastructure? An SMB software distribution platform? Or a leveraged portfolio whose value depends on integration and pricing execution? The honest answer is that the word "hosting" now conceals three increasingly different business models — and the books behind the same word are not comparable without measurement.

1 · Hosting has stopped being one business model

GoDaddy is monetising the funnel. In Q1 2026 its customer count was essentially flat (20.4 million, up just 13 thousand since year-end) while ARPU rose 9% to $246, Applications & Commerce revenue grew 11.6% against 2.8% for the core platform, normalized EBITDA rose 13% and free cash flow 15%. That is a product-mix and monetisation story, not a customer-growth story. It reports Q2 on 30 July.

IONOS is pairing web presence with sovereign European infrastructure. Its Q1 added a record 180 thousand customers (to more than 6.8 million), grew revenue 5.7% and held a 33.9% adjusted-EBITDA margin — while management leads its story with AI and sovereign European cloud, a positioning no US-listed peer can copy.

Europe's consolidators are attempting to become SMB software platforms. team.blue passed €850 million of ARR in 2025 (€866 million reported, 11% organic growth) and now buys e-commerce platforms, compliance tools, marketing attribution and AI products more often than it buys hosters. Your.World raised €800 million of preferred equity from Ares and Carlyle behind a serial-acquirer model. Cinven and Ontario Teachers back group.one on the same thesis: a resilient subscription base plus an M&A pipeline plus the opportunity to sell higher-value products into it.

Public equity, private equity and private credit are converging on the same customer base — but they are underwriting three different machines.

2 · The same word, very different books

Reported customer counts are not comparable across these models, because the underlying books differ materially in activity, geography and product adoption. Measured on one consistent yardstick — confirmed-active business domains, the floor of domains showing live business activity, not raw registrations — the footprints as of 26 July 2026:

Scale is the least interesting difference. Three others matter more to an underwriter:

• Book activity. Of GoDaddy's ~52 million control-plane domains, 35.6% are confirmed-active business — the rest is dominated by parked and idle inventory, the signature of a registrar-led funnel. The European groups run far smaller but denser books: 67–78% of their control-plane domains are confirmed-active. Neither shape is better per se — they are different machines. A funnel model wants a huge low-intent top; a subscription-book model wants density.
• Product adoption. Observed modern-email adoption on GoDaddy's confirmed-active book is 39.6%, versus 8.5–14.3% across the four European groups. That is a DNS-observed signal on near-complete coverage, and it is exactly the monetisation gap the ARPU story runs on — and the cross-sell headroom the European groups are implicitly selling to their sponsors.
• Concentration. The United Internet/IONOS book is 38% .de — a dominant home-market position; team.blue and group.one spread across dozens of national brands with no TLD above ~20%. Geographic concentration cuts both ways: pricing power at home versus diversification across markets.

3 · Europe's consolidators are moving up the stack

The M&A ledger is unambiguous about the strategic shift. Counting only validated acquisitions and mergers (press-confirmed, deal-kind strict — portfolio enumerations and rebrands excluded):

19 of team.blue's 34 validated acquisitions are software, not hosting — sales tools (Saleskit), influencer-marketing platforms (Kolsquare, Woomio, Storyclash), marketing attribution (Windsor.ai), e-commerce platforms (Shoptet, Ticimax) and an AI web-app builder (Macaly). group.one's largest disclosed cheque is a social-media-management SaaS (SocialPilot, US$50M+), not a hoster. cyber_Folks took control of PrestaShop — an open-source e-commerce platform — its boldest non-hosting bet. Your.Online's most recent buys are security, identity and AI products. The hosting roll-up is becoming an SMB-software roll-up.

The critical investor question follows directly: are these groups building a distribution advantage, or buying software revenue at software multiples? Deal count proves the strategy exists. It does not prove cross-sell, product adoption or revenue synergy — those need evidence at the customer-book level: whether acquired products actually address the existing hosting base, whether brands have real access to one another's customers, and whether attach shows up in the observable stack.

4 · Integration is the hidden valuation variable

The divide that will matter next is not large-versus-small. It is integrated-versus-assembled — groups whose acquisitions have converged onto common platforms versus groups that still operate as loosely connected brand collections. HostingBrain measures this from public infrastructure signals: for every ledger-matched acquisition, has the acquired brand converged onto the group's shared hosting, mail and DNS backbones?

Measured on the 26 July snapshot, acquisition integration spans 0.68 (United Internet/IONOS) down to 0.08 (Miss Group) — the widest spread of any variable we track across these groups. United Internet/IONOS runs the closest to one platform. group.one and cyber_Folks sit mid-table. team.blue is the interesting case: 82% of its book sits on group platform hubs, yet its ledger-matched acquisitions score only 0.40 — consistent with a deliberate multi-local-brand strategy in which acquired books keep their own stacks. Your.Online and Miss Group operate, on this measure, closer to brand portfolios than platforms.

Two honest limits. This is an external integration indicator, not a verdict: a group can centralise servers while failing to cross-sell, and it can run excellent commercial operations on deliberately separate infrastructure. And low integration has three candidate explanations — deliberate decentralisation, delayed execution, or technical debt — that the signal alone cannot separate. What it does provide is something reported financials cannot: an independent, repeatable test of the integration story management presents. Groups that claim platform economics should expect this number to move; a holding-company strategy should expect to defend why it doesn't need to.

5 · What finance houses should underwrite

One more measured input belongs in every model. Hosting's stickiness is real but routinely oversimplified: 73.6% of multi-domain owners keep every domain with one group-folded provider, and the average owner concentrates 88.9% of their domains with their primary provider — yet observed provider switching on Europe's confirmed-active book still annualises to 16.2% of domains. These are different populations measured at different grains (owner-level wallet concentration versus domain-level provider movement) and must not be reconciled into one churn number — but together they kill the lazy "negligible churn" line: customers concentrate their wallet, and meaningful movement happens underneath at the domain level. Loyalty, domain retention, revenue retention and infrastructure migration are four different curves.

The diligence checklist that falls out of the three models:

1. Book quality, not book size — confirmed-active share of the reported base, dead-page and idle-inventory rates, geographic concentration.
2. Where growth comes from — organic volume versus price/ARPU versus acquired revenue; GoDaddy's Q1 is the template for reading a mix-led quarter.
3. Renewal economics — the register-to-renew price architecture across the book, market by market.
4. Customer and domain movement — wallet concentration and domain-level switching, measured, not asserted.
5. Integration depth — and whether the trajectory matches the platform story being told.
6. Software attach — evidence that acquired SaaS reaches the hosting base, in the observable stack rather than the deal deck.
7. Sponsor horizon and structure — who backs the group and where they are in their hold.

6 · Five questions management should be able to answer

1. What share of your reported customer base shows live business activity — and how has that share moved over your hold?
2. Of your last ten acquisitions, how many books have migrated onto shared platforms, and what did each migration do to churn?
3. What fraction of hosting customers hold at least one of your acquired software products today, versus at acquisition?
4. How much of revenue growth is price, and what does domain-level movement look like in the quarters after a renewal-price increase?
5. If your integration score is low by design, where does the operating leverage in the platform thesis actually come from?

Reproduce this — or ask it yourself

Every HostingBrain figure here is queryable through the connector. The consolidation map and the software-M&A timeline are one call each; integration scores and the loyalty split are one more. In Claude or any MCP-compatible assistant: