A/B Testing Buy Now Links

The conventional wisdom says that a Lite version of your app draws in more sales to your paid app1. In my experience this has been true. Generally, the Lite version offers a limited experience with an option to upgrade to the full app at a certain point. The Lite version should offer just enough content or functionality to give the user an idea of what the app is like and still leave them wanting more. Getting this balance correct is crucial to the success of the Lite-to-Paid conversion. Other factors also come into play such as the verbiage used at the point of decision.

In order to determine the optimum text I decided to A/B test the text on the link used to launch the user over to the paid version of the app on the AppStore. The button is very visible right on the main screen of the app. Can’t miss it. When the app is first launched it randomly chooses one of four different variations of the text (I know, I know, I should probably call it an A/B/C/D test) and displays that version of the button text to the user. In order to keep things consistent the app tracks which version was selected and continues to show that version to the user from then on. The four variations that I chose were: “Buy, “Buy Now”, “More”, “Get More”.

For tracking purposes I added a signature to my LinkShare link which gets tracked whenever someone clicks on the link. LinkShare not only tracks the click throughs on these links but also purchases made in iTunes which originated from a click on one of them (maybe I will cover this in another blog post if there is enough interest). The system isn’t perfectly accurate since LinkShare tracks ALL sales associated with a link rather than just the sales to the app you linked to. Meaning that if someone clicked over to your app but decided to buy the $150 Beatles Box Set instead then that sale would be attributed to the link signature that was used. However, given a large enough sample space the non-app sales should even out and give a fairly good idea of the relative sales to your app from each version of the button text.

As it turns out I never really did anything with all of this. It’s been in the codebase for a year and a half. Ouch! But, what this does provide is a nice and large sample set to look at. Hopefully, I can put this info to good use at some point. Even better I hope that it is of a more immediate benefit to you.

In the table the ‘Member ID’ column is the custom signature that was set in the LinkShare link (one each for buy, buy now, etc). The clicks are the total number of click throughs (CTR) from the first of June of 2009 until Feb 4, 2011. The Sales column is the total amount of iTunes merchandise purchased as a result of a click on one of the links and Commissions are the LinkShare revenue generated on those sales.

The first thing that stands out are that the buttons with the text ‘More’ and ‘Get More’ generated more than twice the click throughs as the ‘Buy’ and ‘Buy Now’ links. Most likely this is due to the somewhat ambiguous nature of the the terms ‘More’ and ‘Get More’ (unintentional I assure you). I don’t really think the user knows quite what to expect when tapping buttons with those labels and purchasing the app is probably not what many expect. Or maybe it’s curiosity. What’s behind the ‘More’ button?

The ‘Buy’ and ‘Buy Now’ labeled buttons set expectations appropriately. When a user taps on one of them they pretty much know that they are going to a place where a purchase may take place (aka the AppStore). While the raw number of clicks is less than half that of the ‘more’ variety the sales associated with the ‘buy’ labeled buttons confirms this. For instance, for the ‘Buy’ button an average of 0.44 sales occurred per tap (1640/3768 = 0.435). At the same time for the ‘More’ button an average of 0.24 sales occurred per tap (2261/9524 = 0.237).

What does all of this tell us? While the ‘Buy’ links have a much higher conversion rate the ‘More’ links create many more click throughs and have higher raw sales numbers2. I do have to say that to me the ‘More’ style links seem a bit disingenuous in that they do not clearly indicate to the user what is going to happen. So, there you have it grasshopper. With great power comes great responsibility. Use these findings as you see fit.

This post is a part of the iDevBlogADay group of Indie development blogs. Thanks to @mysterycoconut for managing such a great site.

1. Current thinking may slant more in the direction of the Freemium model where the app is free but offers an in-app purchase in order to ‘upgrade’ to the full content.
2. What this also tells us is that Lite versions do work to a certain degree as witnessed in the raw sales numbers across all variations of the label. I will say that the Lite version drove <10% of total paid app sales during that time period however.