Increasing Customer Engagement for a Q-Commerce Brand
2M+
Customers
Across more than 70 countries on four continents.
1
Quarter to alpha
A notably tight timeline for a product this intricate.
8
Vouchers and notifications
Eight of each, with specific messages, driven automatically by the pipeline.
The client
A prominent e-commerce company serving more than two million customers, across over seventy countries on four continents. Its apps and couriers connect customers with shops, supermarkets and restaurants.
The platform pioneers q-commerce and serves several verticals through a range of brands.
The challengeWhat the teams actually needed
The goal was more profit through better engagement and less churn, by interacting with customers in a way that makes them use the platform regularly.
Challenge 01Knowing how to approach someone
Understanding the right way to reach a customer was the precondition for everything else.
Challenge 02Three levers
The client can send messages, dispatch emails or offer incentives such as vouchers and discounts.
Challenge 03Persuasion against irritation
Working out which actions would persuade and which would annoy — and so cause the churn being fought — was the critical judgement.
What we did
A machine learning pipeline suggesting the next best action
The client’s own data scientists built the model; Netguru engineered the software and ran the machine learning operations around it, working as a team extension day to day.
End to end
The Next Best Action pipeline runs from gathering the data through engineering and calculating features, on to the model’s prediction, the reward calculation and the reporting that closes the loop.
Predictions in production
Netguru built the infrastructure for predictions, created the tables in the data warehouses and integrated with the communication channels already in use.
Vouchers sent by the system
Through an API already in place, an automated system issues a voucher or sends a message, choosing between eight vouchers and eight push notifications with particular wording.
New actions, easily added
The system takes new actions readily — a percentage discount on the next order of a given product line, say, or a dedicated voucher with a personalised message — and keeps learning as it runs.
More like thisOther places we scaled product teams for growth.
What’s next