Cases · Grow
How a leading fintech nearly doubled conversion in its cross-sell campaigns
In a live A/B test with the same campaign for both groups, customers selected by Avra's predictions converted 90% more.
- Client
- A leading fintech
- Stage
- Grow
- Measured on
- Live A/B test, same campaign
- Status
- Moving toward production
The challenge
A leading Brazilian fintech gains tens of thousands of new business accounts every month and wants to turn them into POS terminal users. Its campaigns went out to the whole base, with no way to tell which new accounts were likely to buy, resulting in a huge and inefficient acquisition cost.
For more than a year, the team tested Open Finance data, its own transaction signals and third-party data providers to rank these customers. None of them produced a measurable lift in conversion.
What Avra built
Avra built a conversion propensity model trained on the client's history of more than 340,000 business accounts. Each new account gets a prediction as soon as it opens, using only its CNPJ and signals from Avra's foundation model, trained on the graph of Brazilian companies and their relationships.
The model then went through a live test. The client sent the same card machine campaign to two groups: one selected by Avra's predictions and one control group. The only difference between them was who received the message.
Results
Avra's prediction was the first approach the client tested that produced a measurable lift in conversion. Same campaign, same message: the whole gain came from choosing who to target.
- +90%
- conversion rate among the top-ranked half of the base
- +16%
- conversion rate even among the lowest-ranked half
Next
Based on these results, the client is moving the model toward production.
Reach the customers ready to buy.
Avra ranks every new customer the moment they arrive, so your campaigns start with the right audience. Talk to our team to see what it can do for your cross-sell funnel.
Talk to sales