Challenge
A retail group runs more than 700 stores across Europe. Behind every store sits a lease, a set of obligations, a floor plan, the people responsible for it and a stream of dates that cannot be missed: rent reviews, renewals, break clauses, inspections.
All of that lived in contracts, spreadsheets and inboxes, spread across countries and teams. Finding out what a lease actually said meant someone reading it. Deadlines depended on someone remembering them. And when the expansion team asked whether a new location, or an existing one, was worth it, the answer took weeks of pulling information together by hand.
Solution
We built one system to run the whole store estate, a CRM for stores rather than for customers. Every store has a single record holding its contract, its key terms and dates, its floor plan, its performance and the people responsible for it, and the system sends the reminders before a deadline arrives rather than after.
AI does the reading that used to take weeks. It extracts the terms that matter from every contract, in several languages, and analyses floor plans so stores can be compared like for like. Performance data comes in from the stores' point-of-sale systems, space data from the architectural team, and everything that used to sit in scattered spreadsheets is cross-checked and brought into the same record. With the whole estate in one place, the expansion team can judge how profitable a store is, or could be, from the data instead of from a spreadsheet built for the occasion.
Execution
Modelling the store, not the spreadsheet
We started by defining what a store is to the business: the record everything else hangs off. A lease, its clauses and dates, the space and how it is laid out, the people with access and their roles, and how the store performs. Getting that model right first meant every later feature had one place to read from and write to.
Reading contracts at scale
Leases differ by country, landlord and decade, so the extraction was designed to find the same set of terms in every one of them, from rent and indexation to renewal and break dates, and to show the clause each value came from. People confirm what the system extracts, and the confirmed terms drive the reminders automatically.
Reading floor plans
Floor plans arrive as drawings, not data. The system analyses them to pull out what matters for comparing stores, such as usable space and how it is divided, so the expansion team can compare a store in one country with a store in another on the same terms.
Bringing the scattered data together
The numbers that matter for a store lived in different places: sales in the point-of-sale systems, space and layout with the architectural team, and a long tail of facts in spreadsheets kept by different people in different countries. We connected the first two as sources and cross-checked the spreadsheets against them, so conflicting versions were found and resolved once, instead of argued over in every meeting.
Giving every team the view it needs
Store operations, legal, finance and expansion each see the estate through their own lens, with access managed per user and per role. Operations sees what is due this week; the expansion team sees how stores compare and where the next one should go.
