Most product and marketing tests still aim at the moment of purchase. Wingify’s State of Experimentation 2026 report found that 45% of feature experiments in its dataset focus on conversion rate optimization, while 6% focus on retention and engagement and 5% on whether users adopt new features.

What the report says

The report has two main parts. The first is a survey of 108 organizations across 21 industries, all users of the Wingify platform, which rated themselves on 17 capabilities. None scored below 40%, and 36% placed in the top band. Even so, 56% run only one or two experiments a month, and 55% named limited bandwidth and human resources as the barrier to scaling. Of the organizations running one or two tests a month, 75% said increasing test velocity is their top priority for the next year.

The second part analyses feature experiments run by brands in travel, fashion, grocery, media and finance. Wingify, which sells experimentation software, calls retention the “massive, overlooked opportunity” in the data. The report also says 86% of surveyed teams use AI in testing, most often to generate ideas and variations.

Why it matters

A team that runs two tests a month has 24 chances a year to learn something. If the report’s mix holds, about 11 of those go to checkout and sign-up flows and one or two go to what happens after the sale. Acquisition costs are paid again for every customer who does not come back, so the testing calendar is also a budget decision.

What to do with it

In our reading, the self-ratings deserve some caution, since the respondents are Wingify customers scoring their own programmes in a survey run with the vendor’s customer success managers. The test mix is the more useful number. Count your last 12 months of experiments by goal and compare the split with 45%, 6% and 5%. For how to treat vendor figures in a test plan, see our argument on launch-day performance numbers. Loyalty decision-makers already expect budgets to grow, so retention tests should find money behind them.

Source: Wingify, The State of Experimentation 2026