From "Which City?" to Board-Ready: The Complete GrowSmart Tour

An expansion decision is never one question. It's a chain of them: which markets deserve attention at all, what's really in each one, what could kill the case, whether the thesis survives a second opinion, where exactly to land, and how to carry a board with you when you commit. Most tooling answers one link in that chain. GrowSmart was built to run the whole thing.

Here's the full tour, in the order a decision actually moves.

Step 1 — A scored verdict, in minutes

Give the engine any of 208 cities across seven regions and a category, and it returns a complete market-entry analysis: a demand score and an opportunity score (0–100, scored separately, because a market can have enormous demand and no room left in it), a verdict — Strong Go, Conditional Go, Hold or No Go — against published thresholds, the actual named competitors mapped to their locations, the specific supply gaps by capacity band and district, and full unit economics: capex range, lease benchmark, breakeven occupancy, target occupancy, and payback under bull, base and bear.

Every claim is sourced, dated, and confidence-badged — Grounded, Mixed or Estimated — with clickable citations down to the individual numbered source. This is frontier AI at high reasoning effort, grounded in live web data, behind a strict senior-strategist rubric; the London read names Convene at 22 Bishopsgate, not "several established incumbents."

And once any teammate has run a city, it's cached for the whole team — the second look is instant.

Step 2 — Screening at portfolio scale

One verdict is useful. The product really opens up at the pipeline level. The City Intelligence Index puts every scored market in one sortable, filterable table — verdicts, scores, market sizes normalised to USD billions so a GDP in rand ranks honestly against one in dollars — with a world map view and side-by-side city comparison for the finalists.

Because every city runs through one framework, the ranking means something: a difference between two scores is a difference between two markets, not between two analysts. Screening forty candidates stops being a budgeted event and becomes an afternoon.

Step 3 — Deep Dives where the decision needs weight

When a shortlist survives screening, three on-demand Deep Dive reports extend the recommendation exactly where boards push hardest: Risks, Target Segments, and Next Steps. Each is a 400–900 word, memo-style report with structured per-item cards, its own sources, a confidence rating per card — and an explicit list of any claims the research couldn't verify, surfaced rather than smoothed over. That discipline is a feature: fewer verified claims beat more invented ones in front of a committee.

Step 4 — The LLM Council: a second, third and eleventh opinion

For the decisions that matter most, one model's view — however good — is still one view. The LLM Council fans a city out to a panel of frontier models, each producing a full independent analysis you can open and read. The consensus is built the defensible way: numbers consolidated deterministically (median scores, majority-vote verdicts, frequency-ranked competitor lists) before a chairman model writes a word of prose, so the narrative can never move the figures. The consensus report shows member count, verdict agreement, and the min–max spread behind every median.

When the panel converges, you have conviction. When it splits, you have a map of exactly where the market is genuinely uncertain — which is worth more than false confidence.

Step 5 — Calibrated to what you actually build

The framework is constant; the signals inside it swap to your category:

The calibration is not cosmetic. The same 208 cities return near-opposite verdict distributions through a 5-star hotel lens and a boutique lens — because the cost bases differ, and the engine scores the cost base, not "hospitality" in the abstract.

Step 6 — For restaurant groups: concept, rollout, street

The restaurant vertical goes furthest down the chain. Store a concept brief — your cuisine, your check size, your positioning — and every analysis is conditioned on it, scored against the actual named incumbents you'd trade against. From there, a rollout plan sequences your candidate cities into ranked opening waves, with cannibalisation penalties against your own estate and deterministic thresholds locked in code. And inside a winning city, a catchment report goes street-level: five to eight named catchments with sub-scores, rent bands, anchor streets and named competitors — every claim then adversarially fact-checked in a second grounded pass, with anything refuted kept visible, flagged and downgraded.

That's the full arc: from "should our brand be in this city" to "this catchment, in this order, after these two openings."

Step 7 — Board-ready in one click

Every analysis exports as a PDF: a one-page executive summary for the pre-read, or a full report — city overview, market demand, competition with a venue map, location scores, financials, SWOT, recommendation, numbered sources and a "Data as of" date — with cached Deep Dives attachable in the same export. Council members and the consensus export the same way. The output is built to be stapled into an IC paper, not screenshotted from a dashboard.

Underneath it all

Coverage refreshes on a defined cycle, and anything stale or missing is visible as such rather than silently presented as current. Analyses re-run on demand when a market moves. Teams get up to eight seats on Pipeline (£45–100k/yr, all 208 cities, unlimited), and Enterprise adds scoring calibrated to your investment thresholds, a private data environment, your branding — and sector exclusivity across your competitive set, locked to whoever moves first. Investors run the same engine sideways: a comparable demand-durability read across a sponsor's whole pipeline, inside a deal timetable.

The takeaway

Every link in the expansion chain — screen, verify, stress-test, localise, present — currently lives in a different document, from a different firm, on a different definition. GrowSmart's actual product is the chain itself: one engine, one framework, every step, with the working shown at each link.

Book a demo and we'll run the full chain live on a market you're weighing — or start with a paid 1 site/city pilot (five business days, fee credited against an annual plan) and judge the output on a market you know cold.