Free trials · Two-year randomized field experiment with a three-stage conversion model
Longer or shorter? A large-scale randomized field experiment on the impact of free trial duration on sustainable user conversion in the Freemium model
Frontiers in Psychology · 2025
Summary
In one global Freemium SaaS experiment, a seven-day rather than three-day premium trial increased trial adoption and delayed conversion, but did not significantly change immediate conversion.
For your marketing
Possible founder application
Use this as a measurement-design example: separate trial adoption, immediate conversion and delayed conversion, and predefine how product use and promotional treatment will be analyzed. Estimate the effect with a local experiment rather than importing the reported percentages.
Reported result
What the study found
The seven-day trial increased trial adoption by 11.098% and delayed conversion by 42.36%, while the experiment found no statistically significant effect on immediate conversion. The paper also reports that longer-trial users responded more to feature-based promotions, while shorter-trial users were more sensitive to price incentives; effects varied with product functionality and market context.
Research question
What the authors investigated
In a Freemium SaaS setting, how does premium-trial duration affect trial adoption, immediate paid conversion and delayed conversion?
Design
The authors report a two-year randomized field experiment comparing a standard three-day trial with a seven-day trial at an anonymized global image-editing SaaS provider. They use user-level behavior data and a local average treatment effect framework to evaluate a three-stage conversion funnel.
Sample and setting
680,588 new users across 190 countries during 2023–2024 at one leading global image-editing SaaS company. The provider and product are not named.
Scope and transfer
Limitations to keep in view
One anonymized provider and product, one historical period, two trial lengths and a Freemium access model. The reported stage-specific outcomes and promotional interactions may not transfer to another product, audience or trial design. A 2025 correction adds an omitted author affiliation; it does not change the reported findings.
Research sources
The full open-access article was reviewed for the experiment, sample, comparison and findings. The publisher correction was checked and concerns the author’s omitted affiliation.