Analysis: What is the value of returning users (ad or banner-blindness) ?

What is the value of returning users - and should we start reducing ads shown to long-time users as a “reward” (basically because these users are not likely to click on ads anyway because “banner blindness” has set in fully)

It is an irony of this ad/app space that the MAJORITY of the revenue is earned from NEW users - this is especially true for apps which have poor retention. So the worse an app is, the MORE it depends on the new users.

New users ALSO tend to click more on ads - as becomes clear to many developers as they see how new user/returning user percentage seems to be directly related to the revenue for that day. Or whenever they change something in the app - or issue a new app - the revenue per DAU seems to be higher (perhaps because of the higher percentage of new users). It seems new users tend to click all over the place - compared to returning users. In addition, “banner blindness” is all familiar to us - as we tune out ads on web browsers on desktops - though on apps it maybe harder to ignore a flashing banner ad etc.

It would be great if Google Developer Console showed a graph for (new user/returning user) for easier graphical sense.

So some variation in revenue can occur due to variations in new user/returning user as well.

I ran an experiment using Leadbolt HTML AppWall - since this uses a URL which can be changed, I changed the url that was used depending on “how many days since installed”.

1st day - 68% of impressions - 71% of clicks - $0.97 eCPM - $0.03 CPC - 3.76% CTR
2nd day - 5% of impressions - 5% of clicks - $0.17 eCPM - $0.00 CPC - 3.58% CTR
3rd day - 3% of impressions - 3% of clicks - $0.09 eCPM - $0.00 CPC - 3.23% CTR
4th day - 2% of impressions - 2% of clicks - $1.74 eCPM - $0.04 CPC - 4.21% CTR
5th day - 1.5% of impressions - 1.5% of clicks - $1.00 eCPM - $0.03 CPC - 3.74% CTR
6th day+ - 20% of impressions - 17% of clicks - $0.89 eCPM - $0.03 CPC - 3.09% CTR

NOTE: the 6th url includes day 6 and all subsequent days (i.e. including the “long tail” of users). You can see that this is significant i.e. 20% of impressions - though their click-rate is much lower. Conversely, for first day users, the click-rate is higher than expected (because of the newness of the UI factor for new users).

The other variation between days and in CPC/CTR is not that relevant (since it varies between apps). So only look at the very obvious differences (not the subtle ones - as those can be just because of random variation because of low impression numbers).

The patterns which ARE dominant are that:

  • first day has highest impressions (for apps with a lot of returning users this will obviously be much lower for the first day - versus all other days visited)

  • you can see the exponential fall off in users (as they return less often over time) (day 6+ is a sum of day 6, 7, 8 …)

  • the pattern that CPC is near-zero for day 2 and day 3 is similar with other app (so maybe significant) - however CTR is high - so the reason is that the ads were low paying (so how is this in any way correlated ? - unless Leadbolt is offloading low-paying ads to returning users who come back on day 2 or 3 ?)

  • it answers the question “should we keep showing ads since the user interest will drop off over time” - the answer is that for day 6+ (i.e. day 6, 7, … to infinity) the revenue potential is 17% of total revenue or so - which is significant. So EVEN jaded, bored returning users will earn you significant revenue.

NOTE: what this statistic DOES NOT RESOLVE is the question of how revenue goes down by REPEATED return to app - for that ANOTHER experiment would have to be done - where the url changes NOT by the day of return to app (after first install), but by the COUNT of every new day that returned to app. This will help disambiguate whether users who return to app after using app 10 times are equally valuable or not. This statistic could be used to therefore REWARD returning users so after they have used the app 10 times (say) then there are no more ads shown (if your statistic shows that users who have used app more than 10 times are UNLIKELY to ever click on an ad again - i.e. the “banner blindness” is fully active for them).

If you have statistics about how well returning users (those who have used your app more than 5 or 10 times i.e. on 5 or 10 different days - not on the same day) - then let me know.


Good research, but one point you seem to have not addressed is that a lot of networks tend to show the same lists of ads repeatedly, and only getting fresh content once in a while. Not sure if this is the case with Leadbolt, though, and it might also be a case of not many ads being available for my country, causing my perspective to be limited.

I’m not entirely sure how much impact would that make, but if there’s a sizeable portion of your users seeing ads they already clicked in, it’s perfectly logical that your click rate plummets starting at the second day. Would also explain why there’s still as much as 20% for day6+, as if a user uses the app on Day 1 and Day 9, odds are the ad network will be giving them different content the second time.

Google Analytics does this.

Excellent post adforandroidapps!

What we can conclude here is that banner blindness does not seem to exist because the CTR did not drop significantly, right?

Regarding the day 2 and 3 “anomaly” …

Yes, that could be it - however that has not stopped the CTR - it’s just that Leadbolt is paying less - or users click but don’t download as often (i.e. to completion).

However there is a fallacy in this argument (and the one you make about users not clicking as much on repeated showing of the same ads) - and that is that given CTR is 3% or so - the ones who did not click are still an audience that has not clicked yet. So on day 2 or 3, the BULK of those clicking on ads would still be first-time clickers. UNLESS there is some statistic which suggests that it is usually the SAME 3% of the population who are the ad-sensitive variety (i.e. whether it be day 1, 2 or 3, it is the SAME sub-group of people who wind up clicking on ads) - and there may be some truth to that (i.e. the ad-vulnerable public - as opposed to the public which is determined NOT to click on any ad ever).

You are correct, I must have misread it the first time and didn’t notice CTR was staying the same despite the drop in percentual revenue.