I tried this with an app, that * is online since almost 1 year * has over 50.000 users. * has over 30 activities * for long term users, it's used as reference (users come back every day, open one specific activity and leave again, some type of "daily prayer")
Statistics after update with random interstitial (40% of the users have already updated to the new version): * No changes in daily installations (pretty steady before and after) * No changes in daily deinstallations (pretty steady before and after)
Ad-Mob Income (only from interstitial and I compared the same days of the week): No randomness (Jan 16th - Jan 22nd) 13.025 impressions 627 clicks RPM €4.08
I lost over 50% of my income, that is even more than I expected.
I only gave it one week, but the down trend is too obvious to me, to see a need for extenting the trial period.
I am pretty sure, that this is working well for other concepts (games probably), but from what I have seen, I stick with the interstitial upon opening the app.
@sylviathewitch: Nice analysis, thanks for being kind enough to share this useful info! I predicted 20% to 40% loss of revenue, but wow, this is mind blowing! I'm grateful we've been given the option to turn the randomness off.
@turbomike: I'm not sure if you're still here, buddy, but it was an eye-opener for me, when you said, that don't be afraid of using ads, even in paid versions.
@sylviathewitch thanks for posting your results the more information we have the better.
It looks like you had a 33% decrease in the number of interstitial impressions (some drop was to be expected) but a %44 drop in RPM. which is higher than expected. Same goes for your clicks which had a big 65% drop.
Quick question for you if you don't mind, can you give us the same stats for another one of your apps that use AdMob interstitials in the old way over the same time period? I've seen a drop in my apps across the board this month, which is to be expected, but last week was particularly bad. (21% revenue drop across all apps (random and non-random vs the previous week)
Personally in one of my apps that I did not update to the new version of AdMob (I didn't update it at all in fact, you need a control group) I saw the following results:
No Random: Jan 16 2014 - Jan 22 2014 2,173 impressions 56 clicks RPM $4.10
No Random Jan 23 2014 - Jan 29 2014 2,559 impressions 46 clicks RPM $2.17
Which is a 17% increase in impressions, a 17% decrease in clicks, and a 47% drop in RPM. (Ouch!)
For two apps that I did update I saw the following:
(Note: Both apps are still using the original 3-8 random interactions which was decreased to 2-5 in 3.3.1 and may be tweaked as we go forward to include other interactions like list clicks, audio interactions, sharing, saving, etc)
No Random: Jan 16 2014 - Jan 22 2014 Impressions: 7537 Clicks: 183 RPM: 3.11
Random: Jan 21 2014 - Jan 29 2014 (9 days) impressions: 4,606 Clicks: 117 RPM 3.34
Which is a 39% drop in impressions, 36% drop in clicks, and 7% increase in RPM
Another:
No Random: Jan 16 2014 - Jan 22 2014 Impressions: 6089 Clicks: 136 RPM: 2.80 Earnigns: 17
Random: Jan 23 2014 - Jan 29 2014 impressions: 3,698 Clicks: 75 RPM 2.23
Which is a 39% drop in impressions, 36% drop in clicks, and 20% decrease in RPM
So there you have it, those are my stats. You will notice that I saw a 46% decrease in an RPM in an app that I didn't touch. Maybe it was a bad week for the ads served to that app, who knows. If anyone else has some stats I'd love to see them, the more information we get the more we can tweak this.
And of course if you do not like the setting, please do not use it. Adding it as an optional setting was the right thing to do, so thanks for the comments.
Also I've noticed my apps across the board have (random and non-random) have bounced back over the last two days with a 58% increase in revenue compared to the previous two days and that the 23rd was a terrible day.
Also, just to be clear, the idea behind this new ad display code was to increase revenue over the long term. We knew that there would be an initial drop since the ads would not be displayed as much, but we thought that by tweaking when the ads are displayed Andromo developers would see increased engagement from their users and less uninstalls because of ad frequency and repetitiveness, which would compensate for the initial decrease in impressions and eventually lead to more impressions, better ratings, etc.
But this might not be for everyone, if you only have one activity or your users (on average) only see one activity then it might not be the best way for you to monetize your app.
My hope (as the person that wrote the algorithm...please hold your tomatoes) is that we will continue to tweak and perfect this code overtime to come up with the most effective ad code for Andromo developers.
@mark: I'm super tired so just a quick note - thanks for sharing, and yes, my apps also had bigger earnings in the last two days, while before that, the earnings dropped a little, but nothing spectacular.
We also need to have the choice to allow older devices in apps where we're not using any ads. I have users that paid for my apps that now will not be able to use any updated versions.
@Mark here are the stats for a very similar app. Same niche, but only online since 1 month and only 4500 active installs.
Jan 16th - Jan 22nd 2.924 impressions 76 clicks RPM €3.92
Jan 23nd-Jan 29th 3.473 impressions 113 clicks RPM €3.35
And here is another app with a different niche, online since 9 month, 21.000 active installs
Jan 16th - Jan 22nd 19.822 impressions 952 clicks RPM €6.57
Jan 23nd-Jan 29th 19.705 impressions 980 clicks RPM €5.99
So yes, the RPM and the revenue was a bit lower for the second period overall. If I take my whole account (but incl. banner ads) I had a decrease by around 2% for the fillrate, 5% in clicks and about 7% in revenue.
@sylviathewitch Thanks for the stats! Part of what I wanted to show with my example was that even without changing anything I saw a 47% drop in my RPM. So while drops are not good they can happen for a variety of reasons.
But if the feature doesn't' work for you, disable it.
@mark Based on what I can see in Google Analytics, it could be as much as 30%, due to sales through B&N and Amazon. I wonder if the Google Android stats include those devices?
Comments
I tried this with an app, that
* is online since almost 1 year
* has over 50.000 users.
* has over 30 activities
* for long term users, it's used as reference (users come back every day, open one specific activity and leave again, some type of "daily prayer")
Statistics after update with random interstitial (40% of the users have already updated to the new version):
* No changes in daily installations (pretty steady before and after)
* No changes in daily deinstallations (pretty steady before and after)
Ad-Mob Income (only from interstitial and I compared the same days of the week):
No randomness (Jan 16th - Jan 22nd)
13.025 impressions
627 clicks
RPM €4.08
With randomness (Jan 23nd-Jan 29th)
8.704 impressions
217 clicks
RPM €2.26
I lost over 50% of my income, that is even more than I expected.
I only gave it one week, but the down trend is too obvious to me, to see a need for extenting the trial period.
I am pretty sure, that this is working well for other concepts (games probably), but from what I have seen, I stick with the interstitial upon opening the app.
It looks like you had a 33% decrease in the number of interstitial impressions (some drop was to be expected) but a %44 drop in RPM. which is higher than expected. Same goes for your clicks which had a big 65% drop.
Quick question for you if you don't mind, can you give us the same stats for another one of your apps that use AdMob interstitials in the old way over the same time period? I've seen a drop in my apps across the board this month, which is to be expected, but last week was particularly bad. (21% revenue drop across all apps (random and non-random vs the previous week)
Personally in one of my apps that I did not update to the new version of AdMob (I didn't update it at all in fact, you need a control group) I saw the following results:
No Random: Jan 16 2014 - Jan 22 2014
2,173 impressions
56 clicks
RPM $4.10
No Random Jan 23 2014 - Jan 29 2014
2,559 impressions
46 clicks
RPM $2.17
Which is a 17% increase in impressions, a 17% decrease in clicks, and a 47% drop in RPM. (Ouch!)
For two apps that I did update I saw the following:
(Note: Both apps are still using the original 3-8 random interactions which was decreased to 2-5 in 3.3.1 and may be tweaked as we go forward to include other interactions like list clicks, audio interactions, sharing, saving, etc)
No Random: Jan 16 2014 - Jan 22 2014
Impressions: 7537
Clicks: 183
RPM: 3.11
Random: Jan 21 2014 - Jan 29 2014 (9 days)
impressions: 4,606
Clicks: 117
RPM 3.34
Which is a 39% drop in impressions, 36% drop in clicks, and 7% increase in RPM
Another:
No Random: Jan 16 2014 - Jan 22 2014
Impressions: 6089
Clicks: 136
RPM: 2.80
Earnigns: 17
Random: Jan 23 2014 - Jan 29 2014
impressions: 3,698
Clicks: 75
RPM 2.23
Which is a 39% drop in impressions, 36% drop in clicks, and 20% decrease in RPM
So there you have it, those are my stats. You will notice that I saw a 46% decrease in an RPM in an app that I didn't touch. Maybe it was a bad week for the ads served to that app, who knows. If anyone else has some stats I'd love to see them, the more information we get the more we can tweak this.
And of course if you do not like the setting, please do not use it. Adding it as an optional setting was the right thing to do, so thanks for the comments.
Also I've noticed my apps across the board have (random and non-random) have bounced back over the last two days with a 58% increase in revenue compared to the previous two days and that the 23rd was a terrible day.
But this might not be for everyone, if you only have one activity or your users (on average) only see one activity then it might not be the best way for you to monetize your app.
My hope (as the person that wrote the algorithm...please hold your tomatoes) is that we will continue to tweak and perfect this code overtime to come up with the most effective ad code for Andromo developers.
here are the stats for a very similar app.
Same niche, but only online since 1 month and only 4500 active installs.
Jan 16th - Jan 22nd
2.924 impressions
76 clicks
RPM €3.92
Jan 23nd-Jan 29th
3.473 impressions
113 clicks
RPM €3.35
And here is another app with a different niche, online since 9 month, 21.000 active installs
Jan 16th - Jan 22nd
19.822 impressions
952 clicks
RPM €6.57
Jan 23nd-Jan 29th
19.705 impressions
980 clicks
RPM €5.99
So yes, the RPM and the revenue was a bit lower for the second period overall.
If I take my whole account (but incl. banner ads) I had a decrease by around 2% for the fillrate, 5% in clicks and about 7% in revenue.
Thanks again for making this optional!!
But if the feature doesn't' work for you, disable it.