VTB Turns to Machine Learning to Time Its Customer Messages

VTB is rolling out machine-learning models to tailor how it communicates with individual retail customers, adviser Sergei Putyatinsky told the Kazan Digital Week forum.

The system studies a customer's history with the bank: which messages they responded to, when they open the app, and whether they visit branches or bank only online. If someone reliably reads push notifications in the morning and has never answered a call, the CRM system recommends morning pushes instead of a midday phone call.

The bank's CRM already helps pick the most convenient channel - SMS, push, call or email - for a given situation.

By the end of the year VTB plans to extend personalization with new ML solutions built using modern data-analysis methods, including natural-language processing. More than 30,000 features and behavioral characteristics were analyzed for the project.

Putyatinsky said the bank is moving away from standard mass mailings toward a careful, personalized approach to each contact, accounting for channel and timing preferences to make communications less intrusive.

Kazan Digital Week runs October 29-31 and has been held in Kazan since 2020, backed by the Ministry of Digital Development and the government of Tatarstan.