Your edge is not for sale.

Pilots get approved because they are cheap and safe: small budget, no integration, no number attached. That is also why they end. Nothing was measured, so nothing was learned, and the firm concludes the technology isn’t ready when what actually happened is that nobody defined what ready would have looked like.

In practice 95% of AI pilots in real estate deliver no measurable impact — a statistic about governance at least as much as about models.

Ask who is accountable for the metric a project is supposed to move and the answer is usually innovation, transformation, or a committee. None of those people sit on the P&L. A project measured against a number nobody owns cannot succeed and cannot fail; it can only run out of enthusiasm.

In practice If the KPI does not already appear in somebody’s board pack, it is not a KPI. It is a hope.

Off-the-shelf tools encode somebody else’s judgement about how property works, then ask your team to work the way that vendor imagined. It is the wrong shape for a business where the valuable part is precisely the local, unwritten, thirty-year part — the part no vendor has and no subscription confers.

In practice Every firm on the same platform is, by construction, exactly as clever as every other firm on that platform.

Everyone has the data. Land Registry, EPCs, comparables and planning history are available to your competitors on the same terms as to you. What is scarce is the reading: why you passed on that site, what that ward always asks for in committee, whose build costs you quietly halve. That is the asset, and it is currently held in three people’s heads and a decade of email.

In practice When a director retires, the firm does not lose files. It loses the ability to price a certain kind of deal.

Value sits in a handful of decisions and handoffs, and they are almost never where the org chart says they are. Map the process as it actually runs and the expensive part is usually a wait, a re-key, or a judgement call that only one person can make. Buying a tool before you have found that point is buying a faster version of the wrong step.

In practice Rank leverage points by what they are worth, not by how easy they are to automate. Those two lists rarely agree.

The transformation programme is the most expensive way yet devised to discover that an assumption was wrong. Small, reversible and measured beats large, permanent and hopeful — not because ambition is bad, but because the first version is always wrong somewhere, and you want to find out in days on one workflow rather than in quarters across the firm.

In practice One workflow, end to end, in production. A pilot proves a demo works. This proves the business changes.

Not a new metric invented for the project — an existing one, with an existing owner, that was going to be reported anyway. Days to appraisal. Void period. Cost per instruction. If the work cannot be attached to a number of that kind, and to a date, it is a demonstration, and demonstrations should be priced like demonstrations.

In practice A named KPI and a named date, agreed before the first line is written.

Advice that can only ever conclude “hire us” is not advice. The honest answers include an off-the-shelf tool we don’t sell, a process change with no technology in it, and waiting six months until the data is worth pointing something at. A firm that never reaches those conclusions is not being rigorous; it is being paid.

In practice Sometimes the recommended direction is a product we have no commercial relationship with. That is a feature of the method, not an accident.

The work is storing what your people know, not replacing the people who know it. Property runs on relationships and on judgement formed by being wrong in public for twenty years; a system that holds that knowledge makes those people more valuable, not less. If a scope starts reading as a headcount exercise, we are the wrong firm and we will say so early.

In practice The test: after this ships, does the team spend its time on better questions or on fewer of them?