All Latest 617 A/B Tests

Similar to experiment 373, listing descriptions were shortened dynamically using exposable "more" links AND dates/location data was removed. This way, the variation showed shorter listings and therefore more listings per screens. Impact on listing clicks (progression) along with membership starts was measured.

In this experiment, listing descriptions were shortened dynamically using exposable "more" links. This way, the variation showed shorter listings and therefore more listings per screens. Impact on listing clicks (progression) along with membership starts was measured.

This experiment added the simple verb ("play") before the "factory video" label to encourage more video plays. Impact on progression / clicks was measured.

In this experiment, products which were out of stock were removed from listing pages and replaced with in stock ones (not visible in the screenshot). 

This experiment attempted to increase the number of leads on a lead-funnel. As the first step, users were being asked to upload a file. The control showed the file types that were allowed, whereas the variation changed the copy to show a number of benefits for taking that action. The text-based benefits included the: receiving feedback, prices and lead times.

In this experiment, a listing page was expanded to show two actions (apply and view details) instead of a single one (view details only). This variation enabled users with a shortcut action to apply for roles one step earlier (and start membership flows for new users). 

In this experiment, a listing page was expanded to show two actions (apply and view details) instead of a single one (view details only). In the variant, the "view detail" links were replaced with "apply links" starting a job application (and membership flows) sooner. 

In this experiment, the variation showed a listing's owner online status as a badge, instead of showing their average "response time". More specifically, an "Online Now" badge was shown beside individual listings of a 3D printing marketplace site. The experiment measured completed quote / lead requests (a few steps further).

In this experiment, the style of a button leading to view detailed casting calls on a listing page was changed. In the A version the style was a filled high contrast blue background, and the B variation had a feint "ghost button" style. 

In this experiment on a listing page, the search was expanded to show more listings (variation B). Conditionally, if there were fewer than 36 results, set basic filters such as number bedrooms and bathrooms were expanded and appended to the results. Hence if someone chose 2 bedrooms and 2 bathrooms in variation A, they would only see listing with that filter. In variation B however they would first see the filtered results, and later they would also see results with 3 or more of each. 

The core hypothesis of this experiment was that by showing clear availability (in green text) beside each casting call, more users would apply and become premium members. The experiment reports on two metrics: application starts (the first progression metric), and premium membership sales (measured a few steps further in the funnel).

In this experiment, multiple view detail links for a listing tile were turned into higher contrast buttons. 

In this experiment, a list view was tested against a grid view.

The change in this experiment was an exposed SEO panel (B) with a number of clickable filter options. 

In this experiment, two different positions of the View Catalog button were compared. In version A the button was smaller and on the right. In version B the button was wider and more central. (The actual test was inverted before publishing to match the A-B of the pattern.)

In this experiment, a different style for keyword highlighting was used.

In this test we tested onhover buttons (variant) versus more traditional always exposed and visible ones.

In this variation, a number of social proof references were added to a signup modal.

This experiment measured the effect of opening new listing (job applications) in a new tab, against opening them in the same window. The experiment A-B was inversed to match the pattern (in reality, the original already opened the tabs in a new window).

The idea of this test was to add a "Viewed" label on a listing page to indicate listings which have already been viewed by users.