The compounding stack — how home.co.uk, Homemove, and homedata work as one company

We run three businesses across the same vertical. home.co.uk, the property portal. Homemove, the AI-powered transactional layer. homedata.co.uk, the underlying market signal. Most people, when they see all three on a slide, treat them as a portfolio. We do not. We treat them as one company with three customer surfaces, and the architecture matters enough to write down.

This essay is the architecture, and the reason it matters for any founder thinking about building in a vertical the same way.

The shape

The simplest description of the stack is one buyer journey, three commercial surfaces. The same human, in the UK, who is thinking about moving home, ends up touching all three properties over the course of the move — sometimes on the same day. They land on home.co.uk to search for a property. They engage with Homemove when they go to transact. They are influenced, indirectly, by the data on homedata.co.uk through the agents and analysts who feed back into the journey.

The key insight, the one that took us a decade to fully internalise, is that none of the three businesses individually is the most defensible version of itself. They are only durably defensible as a stack. Take home.co.uk on its own — it is one of three property portals with serious traffic in the UK, two of which are larger. Take Homemove on its own — it is one of several conveyancing-orchestration tools competing for share. Take homedata.co.uk on its own — it is one of several property data providers, and not the largest.

Stack them together and the picture changes. The portal generates demand-side intent data. The transactional layer turns that intent into actual deals. The data layer captures the structured signal from those deals, feeds it back into the portal's pricing intelligence, and licenses the aggregated signal externally as a product. Each business is useful on its own and strategic only as part of the stack.

This is the architectural shape we think more founders should be building toward in 2026.

Why this shape is suddenly possible

In 2008 or 2018, this kind of vertical integration was prohibitively expensive. You needed three teams, three product organisations, three sales motions, three sets of compliance. The cost of running three companies as one was roughly 2.7x the cost of running one company. The compounding return rarely justified it for anyone smaller than a Rightmove-scale operation.

In 2026 the economics have shifted in three specific ways.

First, the production cost of any one of the three has collapsed. AI-native tooling means a small team can ship and operate the transactional layer at a fraction of the cost it took to build the equivalent in 2018. Same for the data product. The portal is harder because of legacy SEO and brand investment, but even there the maintenance cost has fallen sharply.

Second, the data plumbing between them is now genuinely cheap. Modern model-based extraction can convert messy unstructured signal from the portal into structured records the data layer can ingest, in real time, at fractions of a penny per record. The cost of "wiring" the stack together has dropped to nearly zero.

Third, the operating team scales sublinearly. A team that knows the vertical can run all three properties more efficiently than three teams that know one each. The compounding context is itself a productivity multiplier we did not have access to before. The same Slack channel that triages issues in Homemove also has the institutional memory of how the portal works.

The result is that vertical integration is now economically rational at a much smaller scale than it has ever been. We expect to see this pattern repeat across other UK verticals over the next three years.

What this means for founders we incubate

The strategic implication for early-stage founders in 2026 is that the right unit of strategic thinking is not "the product" but "the stack." Even at the earliest stage, the question to ask is: what are the adjacent customer surfaces this product touches, and what would the stack look like in five years?

In practice for a typical Moonlabs Incubator company, this might mean:

• The product they are shipping today is the transactional surface.
• The data they are accumulating is, in three years, a licensable data product.
• The buyer attention they are capturing is, in five years, a media-and-distribution surface.

We do not push founders to build all three at once. We push them to be aware that the three exist, to take small structural decisions today that preserve the option to build out the stack later, and to not give away the components that will compound. Specifically: keep the data. Keep the brand. Keep the cap table tidy enough that you can do M&A in either direction in three years. These are cheap to maintain at the beginning and expensive to retrofit later.

The founders who treat the early product as the whole company are leaving the strategic upside on the table. The founders who see the stack from week one make smaller, better decisions across two years that compound dramatically.

What this means for the Incubator's selection

Specifically because the stack thinking is what we know, the Incubator selects more aggressively for founders who can see the stack. We say no to ideas that are genuinely point products, with no path to a second surface and no data flywheel. We say yes to ideas that, even at the earliest stage, sit somewhere a stack could grow around.

This is also why we sit on the cap table at 1% for a long time. The compounding from the stack takes years to play out. Our economics depend on holding through that compounding, not on flipping in eighteen months.

The lesson we keep relearning

The lesson we have relearned, in different forms, every two years across home.co.uk + Homemove + homedata, is that patience plus architecture is the actual compounding instrument. Patience without architecture produces a slow point product. Architecture without patience produces a half-built stack that never gets the compounding return. You need both.

The architecture is what we now teach the Incubator. The patience is what we test for in selection. Both matter. Neither is optional.

If you are a UK founder building inside a vertical right now, and you want to think out loud about what the stack version of your company could look like in three years, we are interested in that conversation. It is exactly the kind of conversation the Incubator was built for.

home.co.uk has run since 1995 and was acquired in 2025. Homemove operates as the AI-powered transactional layer of the property journey. homedata.co.uk is the underlying market signal product. All three are run by James Freestone and Louis O'Connell-Bristow.