Can an expansion strategy be justified rather than intuited?
An international expansion strategy for an art consultancy that cannot be named — three markets, three postures, one justified shortlist. The findings are the client’s; the architecture that produced them is shown in full, redactions included.
The brief
Where the market actually is
The starting point was the market itself. Global art sales concentrate almost three quarters of their value in three national markets — and the client already operates in one of them.
A single stacked bar of global art-market share by value for 2023: United States 42 percent, China 19 percent, United Kingdom 17 percent, and the rest of the world 22 percent — so three national markets hold roughly three quarters of sales.
≈ three quarters of global sales · three national markets
The market evidence resolves the brief into the US and China as targets, with the UK home market as a strand of its own: three geographies presenting three different strategic problems.
Sources: Art Basel & UBS — The Art Market 2024
Three strands, three postures
The architecture of the problem
United Kingdom
DEEPEN
Already operational — the question is not whether but where next within it.
Planning and development data read as a forward pipeline of buildings: every relevant development is a prospective client, found before the hoarding goes up.
United States
NARROW
Legible but vast — fifty states cannot be entered; five cities can.
Census and economic data drive a state → city funnel, scored on a composite index of commercial demand and artistic supply. The output is a justified shortlist, not an intuited one.
China
MEDIATE
Unfamiliar and without a base — direct entry carries the highest risk of the three.
A gateway thesis: two compact, affluent localities — referred to here as Gateway 1 and Gateway 2 — connected to both Chinese and international markets, validated against both territories’ government statistical databases. Their identities are withheld.
DEEPEN · United Kingdom — the lead engine
The client’s existing practice was literal: visit construction sites, read the hoardings, find out who is building. The proposition formalises that instinct — public planning data is a forward pipeline of buildings, and every relevant new building is a prospective client. Applications feed a development pipeline; a use-class filter sorts it by segment; what emerges is a systematic, current, citywide prospect list. The machine is shown below. Its output is live commercial intelligence, and stays with the client.
Sources: Planning London Datahub
NARROW · United States — the funnel
Fifty states cannot be entered; five cities can. Census and economic data drive a state → city funnel, scored on a composite index that weighs commercial demand against artistic supply. The index’s construction is public method; its weights and its answer are the client’s.
A scatter plot of candidate US cities scored on two axes: corporate demand on the horizontal, artistic supply on the vertical. 46 candidate cities are shown at illustrative positions; the 5 with the highest combined standing — upper right of the field — are ringed as the cities taken forward. Their identities and real scores are withheld.
| Index dimension | Weight | US-1 | US-2 | US-3 | US-4 | US-5 |
|---|---|---|---|---|---|---|
| Commercial activityestablishment density · payroll | ||||||
| Corporate presenceheadquarters concentration | ||||||
| Hospitality economyaccommodation & food services | ||||||
| Artistic supplyartist occupation share | ||||||
| Cultural capabilityinstitutions · galleries | ||||||
| Composite rank | — | 1 | 2 | 3 | 4 | 5 |
The candidate field on the two headline dimensions — corporate demand against artistic supply — with the five taken forward ringed. The scorecard beneath names every dimension and its source; point positions are illustrative, and the rank row is the one scored output published.
Sources: US Census Bureau — American Community Survey, US Census Bureau — County Business Patterns
MEDIATE · China — the gateways
The China strand is an argument, not a funnel. Entering a large unfamiliar market directly, with no operational base, is the risk the strand is built to avoid. The proposition routes it instead through two gateway localities — compact, highly affluent, institutionally legible, and connected to both Chinese and international markets — validated against both territories’ government statistical databases. They appear here only as Gateway 1 and Gateway 2. Smaller localities are entered first, and the wider market is approached on its own terms.
A world map showing the real national outlines of the United Kingdom, United States and China. London is marked as a single true point in the UK, the existing base. Five entry cities appear as a schematic cluster inside the US outline, their real positions withheld. Two gateways, labelled only Gateway 1 and Gateway 2, sit at schematic positions to the north-west and south-east of China, with dashed routes leading to a mainland that is never entered directly.
The two gateway localities were selected for their affluence, proximity to Chinese culture and integration with western and international markets. Analysis compared rates of corporate formation with the presence of organisations matching the client’s services to gauge the scale of opportunity, then combined those signals with measures of cultural participation to validate interest in art rental across the client’s active sectors.
Sources: Natural Earth — country boundaries, Government statistical databases — gateway territories
The sweep
24 autonomous analyses
The sweep covers all 8 cities and 3 segments: 24 agentic search runs against a human-set schema, each returning approximately 10 firms. The result is a global register of ≈200 competitors and an independent sense-check on every local market taken forward.
| City | Corporate art rental | Hospitality | Public realm |
|---|---|---|---|
| London | 10 | 10 | 10 |
| US-1 | 10 | 10 | 10 |
| US-2 | 10 | 10 | 10 |
| US-3 | 10 | 10 | 10 |
| US-4 | 10 | 10 | 10 |
| US-5 | 10 | 10 | 10 |
| Gateway 1 | 10 | 10 | 10 |
| Gateway 2 | 10 | 10 | 10 |
24 runs · 8 cities × 3 segments · ≈200 distinct firms after de-duplication
Each run targets the top ten firms in its market and segment, so the counts sit at ten by design. City and segment are structure; the firm names are the deliverable, and the deliverable is the client’s.
Sources: Perplexity-driven agentic search — 24 autonomous runs, human-set schema
Findings, limits and repeatability
What the engagement shows
The market determined the structure of the strategy. Concentrated global sales made the UK, US and China the relevant fields, but the evidence did not support treating them as one generic expansion problem. Existing operations called for a deeper lead pipeline; the scale of the US called for a narrowing funnel; and the conditions around China called for mediated entry.
Each posture became an operational instrument. Planning records turn future development into a renewable UK prospect list, public economic measures make the US shortlist auditable, and territorial evidence supports a gateway thesis rather than unsupported direct entry. The competitor sweep then tests all three strands through one common research schema.
What remains repeatable
The lead engine can be refreshed as planning records change; the funnel can be rerun as market conditions move; and the sweep can update a structured competitor register. The value is therefore not only the shortlist bought once, but a governed way of answering the question again.
What the page withholds
CR-2026-07-17The client is not named, and the recommendations produced for them are not published — that was the engagement’s condition, and honouring it is part of the work. Withheld here means absent: a redacted value never enters this page’s data, its markup or anything the browser receives, so there is nothing behind a bar to recover.
† Every bar on this page stands for a value that was actually produced during the engagement and deliberately withheld — none is decorative. What may be shown is governed by the clearance register whose revision is printed above.
What this work demonstrates
A commissioned expansion study built from international public data: a planning-led pipeline for the UK, a census-driven index funnel narrowing fifty US states to five entry cities, and a gateway thesis for the China strand. The recommendations belong to the client and stay withheld; the architecture that produced them is published in full.
Where this approach could be applied
The same method could support organisations comparing markets, cities or regions where the decision depends on assembling public evidence rather than commissioning primary research, and where the comparison has to be defensible to a board.
Continue the investigation