AI sales recommendations for a spirits group, in 11 markets

Selling in a country means knowing its outlets, and they never stand still: bars and shops open and close every day. Which should a sales rep visit this week, and what should they sell there?

A global wines and spirits group had the answer in an algorithm that recommends, for each rep, where to go and what to sell, with objectives and activities in order of priority. With our applications, a market’s business planners set the strategy, and their sales reps carry it out from the recommendations on their phone, tablet or computer.

What was in place

A spirits group rarely sells to the bar or the shop. It sells to distributors and wholesalers, and they sell to the outlets. So it bought what it could not see: territory data from data providers, and sales data from its distributors.

The data was there, and soon the algorithm too. Nothing put it in a sales rep’s hands. And nothing let the group apply a strategy to a territory: to launch a new product, say, with objectives for where it should be sold first.

The objectives

Deploy new markets fast, then improve the recommendations and the experience from each market’s feedback.
Let business planners apply a business strategy to their territory.
Support sales reps in their objectives, with the best possible experience of the recommendations.
Get results from each market sooner.

What we did

A team inside the programme. We worked day to day among data scientists, data engineers, data analysts, consultants and business teams. We led the applications with a team in France and the United States. We also trained many of them on the platform.

Control towers. For the business planners of each market: decide which products to push and with what weekly and monthly objectives, how many outlets must carry them, which chains to leave out; then launch a run of the algorithm, compare the results, and release one to the field.

Field applications. For the reps: the outlets to visit, what to sell in each, the objectives, and what was done.

Dedicated to each market. Markets differ in their data, their maturity, and the device their reps carry: a phone, a tablet or a desktop. So each has its own applications, more than 50 so far, on shared modules so the next market starts from the last.

The result

Each market’s planners now turn their own strategy into a run and send it to the field. We deploy a market in two to three months, and the programme gets each market’s results sooner: the reps follow 85% of what the AI recommends, and sales are up by 1.5% to 4.5% depending on the market.