Owner Slide In – Inventory Pages

Description / Hypothesis

Placing a feature on the inventory pages that slides in to help owners quickly get back on track once they've gone into the search experience. This feature can be tested for many different callouts/CTAs as a less-obtrusive form of popup.

Variations

Post-Mortem Analysis

As expected, this experiment worked to reduce the number of people getting lost among inventory search results, and improved Owner Sell conversions. Final results saw Variant 2 of this experiment amounting to a 32.3% increase in conversion rate, a substantial gain given the traffic entering these inventory pages.

However, the results of this test were not so black-and-white. Variant 1 finished with an Owner Sell CVR of 0.4%, just slightly behind that of the winning overall variation. It’s also worth noting that Variant 1 perfomed better among Organic visitors (.36% vs. .35%).

Paid visitors told a different tale. On the whole, Paid traffic preferred the control, 0.67% vs. 0.56% (Variant 2). However, when separating by device type, we see this result varies largely by platform. The test was a 30% loss among Paid Desktop users, but in the crucial segment of Paid Mobile users, Variant 2 won out by 140%. Thus, we cannot chalk up a total loss among Paid users.

There is also the impact of other lead types to consider. Variant 1 was demonstrably better at lifting Market Value CVR (30% lift), though that goal type wasn’t this experiment’s primary objective. And, curiously enough, Variant 1 also lifted Ad-Buyer CVR, even though it deployed a strictly owner-focused “sell timeshare” message. CVR increased by 5% (and while it would take further inspection to determine just how this happened, my working theory at this point is that something about the messaging in V1 added a sense of legitimacy that ultimately helped interested buyers).

Invariably, this is an experiment that will require further testing. Like the utility bar, we will need to fine-tune this idea in order to find the optimal deployment strategy for each channel/device type. In light of this, even though this experiment was roundly a win, I will not submit this iteration for implementation, opting instead to continue experimenting with both variants.