Search

How Boost statistics are calculated in Product Discovery

Boost helps you influence which products are more prominently displayed across listings, but understanding how Voyado Product Discovery reports on boost performance is equally important. This article explains how boost statistics are calculated.

What is counted toward boost statistics?

The Estimated Effect measures the relative increase or decrease in session activity caused by the boost. It compares:

  • Boosted group sessions: where the visitor saw products with an active boost
  • Control group sessions: where the visitor qualified for the boost (e.g., correct market or audience), but was not included in the active boost group

Metrics such as clicks, displays, purchases, revenue, and profit are tracked across both groups to estimate the impact of the boost.

The control group is still in the same context as the boosted group, for example, the same page or placement where the boost is configured. The only difference is that these visitors are part of the control group (typically 20%) that the product discovery engine uses for comparison, not the boosted group (typically 80%). This allows the engine to isolate the effect of the boost within the same conditions, while comparing sessions with and without the boost applied. Sessions that interact with a product that is in the boost, but outside of the context, e.g. if the boost is limited to a particular page, but the session interacts with a product via a search result, are not considered at all.

Formula used

Estimated Effect =
(Activity in all sessions / Total number of sessions) ÷
(Activity in control group sessions / Number of control group sessions) − 1

This formula expresses the percentage change in activity between the two groups.

Example

Let’s say a boost action produced the following:

  • Control group: 5,000 sessions generated 10,000 clicks
  • Boosted group: 20,000 sessions generated 90,000 clicks

The calculation would look like this:

  • Average clicks per session (control group): 10,000 ÷ 5,000 = 2
  • Average clicks per session (total): (10,000 + 90,000) ÷ (5,000 + 20,000) = 4
  • Estimated Effect: 4 ÷ 2 − 1 = 1.0, or +100%

This means the number of clicks increased by 100% as a result of the boost.

Data requirements for calculation

For metrics like Sold Units, Revenue, and Profit, at least 500 sold units are required in the control group to calculate the Estimated Effect. This ensures statistical reliability. If this threshold is not met, the corresponding metric will be highlighted in yellow in the interface.

Q: If I boost a product, does all activity with that product count toward the boost?

A: No. Only activity that occurs during the boost period and within the intended session context should count.

Was this article helpful?

/