Research programme
We don't specialise in an industry. We specialise in changing one.
Our research programme takes one sector at a time. We tear it down with the same method every time, publish what we find, and work with the firms best placed to act on it.
Programme status
Sectors
| Sector | Status | Published | Outputs |
|---|---|---|---|
| Home energy & heating | Teardown published | 3 November 2026 | Industry teardown, two insights, 47 public reviews and forum posts analysed, a third-party price survey of 34 Dublin installers reviewed, working prototype of the fix. |
| Next sector | Screening | – | Selected against the criteria below. Propose a sector. |
Selection
How we choose a sector
We score each candidate from 1 to 5 on five criteria and enter where the evidence is strongest.
A fragmented mid-market
Many firms of meaningful size, often owner-run, where one decision maker can say yes.
Public pain
Buyers describe the same failure in reviews and forums, in their own words.
An office-work leak
The failure is in quoting, booking, follow-up or administration, which software can now do, rather than in the craft itself.
Visible economics
Prices, margins and volumes can be estimated from public data.
Answer-layer exposure
Buyers now ask AI assistants who to use and what a fair price is.
Method
What we look at in every teardown
The buyer's moment
When does the customer decide, who is in the room, and what do they need in that minute?
The public evidence
Reviews, forums and complaints in customers' own words, coded for the failure that repeats.
The price and the promise
What competitors publish, how fast they reply, and what they guarantee.
The answer layer
What ChatGPT, Gemini, Perplexity and Google say when buyers ask, and who gets named.
The economics
Unit margins, capacity, regulation and who in the chain captures value today.
The fix and the partner
The smallest AI-native system that closes the gap, and the firm best placed to run it.
Research standards
How we treat evidence
- Quotes are verbatim and anonymised. We keep the writer's own words and spelling, and remove the names of firms and staff.
- Every figure is dated and linked. We record when each source was read.
- Assumptions are labelled. Worked examples say which numbers are illustrations.
- No firm is named as a client unless it has agreed in writing.