12 new regional offices in 18 months. £3.4M attributable revenue. Zero gut-call decisions.

Cardew & Halsey is a UK chartered accountancy network with 14 regional offices serving SME and mid-market clients. The board had committed to opening 10-15 new offices over the medium term — but the methodology for picking which cities was structurally weak.

Each new office cost real money: lead partner relocation or recruitment, regional marketing spend, professional indemnity scaling, and 18-24 months before the office hit fee-revenue self-sufficiency. Get the city wrong and the office bled capital for years. Get it right and it became a regional anchor.

Decision-making was partner-led: a senior would pick 1-2 candidate cities per year, validate via 2-3 partner-network conversations, present to the board, and commit. The data backing each choice was thin — chiefly the senior's intuition about regional market dynamics + a few demographic indicators (population, GDP per capita). Real signals — SME density, FTSE-250 HQ presence, fee-earner migration patterns, professional services concentration, competitor practice closures — were not part of the methodology.

Three of the last seven new offices had failed to hit revenue targets after two years. The board wanted to systematise the decision before authorising the next batch of openings.

The scoring engine indexes every UK region on the signals that actually predict office success: SME density per postcode (Companies House data, refreshed monthly), FTSE-250 HQ proximity (proxy for high-value advisory demand), professional-services workforce concentration (ONS occupational data), accounting practice consolidation activity (ICAEW + ACCA register changes), HMRC-published practice closure data (surfaces under-served regional pockets), and fee-earner migration patterns (LinkedIn data via the firm's existing recruitment platform).

Each input is weighted by a learned coefficient calibrated against Cardew & Halsey's existing 14-office performance — what actually correlated with high-revenue offices historically? Three signals dominated: SME density × FTSE-250 proximity, accounting-practice closure activity in the prior 18 months, and fee-earner migration in-flow rate. The model surfaces those signals explicitly so partners can interrogate the ranking, not just receive it.

Partners get a sorted list of 100+ UK regions every quarter with the underlying scores, the comparable-office benchmarks (Cardew offices in similar profiles), and a confidence band. The board picks from the top of the list. Local intelligence still layers on top (partner relationships, regulatory nuance) but the shortlist is no longer subjective.

12 offices opened in 18 months against the model's recommendations. £3.4M in first-year attributable fee revenue. The compounding loop is now live: every new office's actual performance feeds back into the scoring weights — the model gets sharper, not staler.

12 new offices opened in 18 months. All algorithmically pre-selected from the model's top 25. £3.4M in attributable first-year fee revenue against the £1.8M revenue produced by the prior comparable period (when expansion was partner-led at 1-2 cities/yr).

Office-success rate is the under-the-line number that matters most. Of the 12 new openings, 10 are tracking ahead of plan at the 12-month mark; the remaining two are on plan. Compared to the historical 3-in-7 fail rate, the algorithmic shortlist is delivering materially better office economics.

The board has shifted from picking 1-2 cities per year to evaluating 6-8 candidates per quarter. The cost of evaluation went down (ranked shortlist is automatic) and the quality of evaluation went up (objective signals + comparable benchmarks). The firm's medium-term plan is now to open 18-22 offices over the next three years instead of 10-15 — confidence has unlocked ambition.

Week 1–3: Discovery, methodology audit, signal selection, calibration data from existing 14-office performance. Week 4–6: Companies House + ICAEW + ACCA + LinkedIn integrations, scoring engine, regulatory overlay. Week 7–8: Comparable-office benchmarks, partner dashboard, first ranking validated against the firm's gut-feel intuition (correlated where it should, disagreed where it should). Week 9–11: Production deploy, board training, first official ranking-driven opening committed.