Klaviyo
AI / ML EngineerAppliedAug 6, 2026

Recalibrate purchase predictions for high-return merchants

Predictions can look stable overall while failing for merchants with different commerce realities.

A useful model must absorb event variation without turning temporary data noise into customer targeting changes.

The audience estimate says we have plenty of likely buyers, but the last two sends didn’t behave that way.

Anna Lebedev · CRM Director

He manages lifecycle audiences for an apparel retailer where returns are a normal part of the purchase journey.

What pulls against what

  • cohort accuracy vs. audience stability
  • integration flexibility vs. feature consistency
  • peak-season speed vs. validation depth

What is at stake

Reliable probabilities help high-return merchants plan relevant campaigns without inflating likely-buyer audiences

Why Klaviyo

At Klaviyo, differences in how commerce events arrive can shape the reliability of lifecycle decisions.

Written for

Applied ML engineerData-centric model builderExperimentation-minded engineer

This is the setup. The work is inside.

Running it puts you in the room: the full situation and its constraints, stakeholders who push back in their own words, and the decisions that are yours to make. What you produce becomes a Day One Plan — work you can show someone instead of describing.