Skip to main content
All CollectionsA/B Testing & Optimization FAQ
How does the attribution work in A/B testing?
How does the attribution work in A/B testing?
Lari Lehtonen avatar
Written by Lari Lehtonen
Updated over a week ago

A user is bucketed in a variation A or B or N at the beginning of a visit and if she or he is exposed to one or multiple campaigns that belong to a given variation, even if it leads to a bounce, her or his session is analyzed & reported in reporting view for the relevant variation.

Nosto remembers how users are bucketed and will be provided with the same experience of variation A or B or N in any upcoming session. 


  • On the 1st of February, one user visits an online store with a Nosto test running. He is bucketed in variation B and served by the relevant campaigns of that variation when she or he visit a page type where one campaign of a variation is displayed. She or he leaves and doesn't purchase. 

  • On the 4th of February, user revisits the website and is bucketed to variation B. She or he is exposed to the campaign(s) of the variation B and convert. Nosto reports the conversion corresponding to the variation B. It is a converted visit. 

Note that if in the second visit, the user wouldn't have been exposed to any campaign of the variation B, it wouldn't be stated as a converted visit. Indeed, the relationship between the experience and the conversion would not be strong enough.

Campaign Reach & Attribution 

It's quite common that within a session, the campaign(s) of a variation are displayed multiple times for a single user. This is particularly the case when users are visiting multiple times one same page type. Nosto tracks the performance of the campaigns and the performance of the variations differently. 

If any user is exposed to at least one campaign of a variation and convert, it counts as 1 converted visit for the variation. However, if she or he has viewed 4 times the campaign of a variation, the campaign reach is 4 and will be reported as such on the product recommendation campaign listing view in Nosto.

If one product recommendation or content campaign is used in context of a test as part of a variation while simultaneously used in different placement(s) targeting other segment(s), aggregated statistics will be displayed in the dedicated campaign listing views. However, variation specific statistics will be treated separately in the Nosto A/B Testing & Optimization reporting view.


One product recommendation campaign of Best Sellers (identified by a unique slot ID) is part of a campaign test, targeting Segment of First-time Visitors. This campaign is known as Variation B in context of this test and is displayed in a Placement 123 on the Front Page. 

In parallel, this campaign is also used to target another segment, the segment of Loyal Customers this time, in Placement 456 on Product Pages.

Any First-time Visitor bucketed in Variation B who will be exposed to the campaign due to a visit of of the Front Page, will see her or his clicks reported as part of the variation and her or his purchases reported to the Variation B. In parallel, those clicks are also reported to the campaign-specific performance.

Any Loyal Customer visiting Product Pages who will be exposed to the campaign, who will click one of the recommended products and who will purchase one of these products in the same visit, will be reported from a Click-through Rate, Conversion Rate and Sales on a campaign level only. Indeed, none of these users were part of the A/B Test due to the segment targeting and variation bucketing.

For more information, learn more about placement guidelines here and sales attribution by product here.

Campaign & Variation Revenue Attribution

  • Attributed revenue for Nosto Onsite Product Recommendations follows a same-session Click & Buy attribution method and all revenue displayed on a campaign level is based on that attribution method.

  • Attributed revenue for Nosto Onsite Content Personalization campaigns follows both a View & Buy and Click & Buy attribution method and all revenue displayed on a campaign level is based on any of these attribution methods, based on user selection in the Nosto Admin Panel. Indeed, some content campaigns don't include clickable links.

  • Attributed revenue for any Nosto A/B test variation follows a same-session View & Buy attribution model. 

For example, if a user is exposed to an Onsite Product Recommendation campaign "You may also like" in a visit (say this campaign consists of being a Variation B in a given test), don't click on the recommended products and purchases within her or his visit, the attributed revenue is listed for the Variation B of the test but not under the Onsite Product Recommendation campaign. 

Did this answer your question?