Deciding which retail outlets to send their limited auditing staff to in a given month was not easy for one of our clients. They work with over 10,000 stores in the whole of the UK and want to send an employee there from time to time in order to check the stores’ conditions.

Factors such as location, size and sales metrics could be used to decide whether a store should be audited or not.

With many different stores and many attributes per store, it is difficult to decide which store should be visited next, especially if the business reasons for a visit change week-to-week or month-to-month.

We reviewed products in the market such as IBM Watson Tradeoff Analytics & MIDACO-SOLVER and determined what we would implement differently in order to provide a solution tailored for our client’s needs.


Our application shows the user a map with the individual store’s location on it and a table with information about each store.

The user can then apply filters  to pre-select a portion of the possible stores and choose, if desired, several optimisation criteria that they want the stores to be ranked by.

The necessary information is then sent to the optimisation algorithm that calculates the priority of visiting each store by iteratively computing the pareto-front until all stores are ranked.

The user can also choose to display stores within a polygon with each corner/vertex representing a quantifiable store feature. The location of the store within the polygon shows how balanced its features are in comparison to the rest of the stores and we use an icon to present specific information about each store.

Our solution provides the user with a flexible way of organising the data by filtering and optimisation and also with a good representation to visualize the outcome. This way, they can get a better overview of the data and subsequently derive well-informed decisions.

This solution was delivered in 3 weeks.