Switched interstitials from Admob Interstitials to Leadbolt HTML AppWall (since I can switch them using AppBrain Remote Settings) for a day or so.
I lost $7 in revenue from Admob - and gained $0.25 from Leadbolt.
Are all the non-Admob ad networks in a heap of trouble or what ? And this is not push ads or something - just simple AppWall (leadbolt) which used to deliver $0.03 EPC usually (it’s giving close to $0.00 EPC now).
Should one use their Leadbolt SDK now ? Because I am using their Leadbolt HTML AppWall.
Since each “experiment” causes you to lose cash - I just lost $7 or so for this small experiment - which may not be much but it is 30% of revenue for the day. These losses can add up.
For this reason Admob is fast becoming the default “no-brains” approach - if you choose Admob you cannot be that far off the optimal revenue - and you avoid losing revenue doing all these “experiments”. This is just with Leadbolt - talk about all the PlayHaven/AppFlood etc. type of experiments (and this is just for the AppWall format). All have been disastrous in recent days.
Admob also seems to have successfully converted all banner ads “on-the-fly” to interstitial format - as I suggested in an earlier thread - and this may explain their better ad inventory - though Admob Interstitial fill rate is perhaps closer to 54% (and not exactly 100% as is the case with Admob banners - so perhaps that theory may not be fully correct - i.e. that Admob can convert any banner ad to interstitial format and thus should not have issue with fill rate etc.).
Here are the numbers I got for Leadbolt (may not be a full 24 hours - but this $0.25 revenue came at a loss of $7 at Admob !!!):
Did you try the Leadbolt SDK - or were you also using the Leadbolt HTML AppWall (which has you showing the URL in a WebView yourself) ?
The results I provided are for the URL (I am not using the Leadbolt SDK at all).
At this rate, these fairly stable and reliable ad networks are going to become less and less relevant for developers - as when you lose a developer they don’t come back (though if you are a big app development company - you can be testing all ad companies and switching quickly since that is minimum manpower (to be constantly assessing ad networks) compared to all the other activity at the company). That kind of effort is prohibitive for individual developers.
Leadbolt has both CPC and CPI models in their ads as far as I know. 74 clicks for 1194 impressions is a bit above average. You probably didn’t get any installs out of those 74 clicks which is also normal. I am having conversion rates of 4% across my games. If you got 1 install out of the 74 clicks that would be a conversion rate of 1.35%.
I believe you need more data before you can make assumptions. So maybe run those ads for a few more days if you’re not too worried about losing $7 per day.
Well for equivalent run in “the good old days” (i.e. about few weeks ago) - I would have generated revenue similar to what was lost from Admob.
The stats were sufficient to show up as a drop on Admob but not register equivalent rise in revenue on Leadbolt (now it could be that they are counting the days differently) - however the impression count visible on leadbolt suggests the impressions/clicks were going there without revenue.
It could be Leadbolt has some “no revenue on first day that ads are shown” or something (i.e. triggers some “anti-manipulation” logic in their system). If so, this creates a “hysteresis” into the equation - and makes it harder for developers to evaluate an ad network by just trying them for 24 hours.
However, at this stage I do not feel especially sanguine about continuing the experiment (“bird in hand is worth 2 in the bush”). I had been giving about 10% of the interstitial ad traffic to Leadbolt (so I could “monitor” them). Yesterday there was a slight rise, so I had decided to try it (perhaps Leadbolt had improved and the 10% was not doing it justice). But results (even for not huge impression count) that if there was such improvement it would have been apparent in the 24 hour test.
I wonder if some developers can attest to better performance when using the Leadbolt SDK - I am not using that, but only using the Leadbolt HTML AppWall where you show a URL in a Webview).
By better I mean something above $1-$2 eCPM for Leadbolt AppWall.
Same for PlayHaven/AppFlood - if someone has seen performance better than Admob there (I have yet to hear a single good story for those ad networks - though they ARE used in a lot of games). Though it could be that they are so great that developers choose to not mention them glowingly in case developers rush to use that ad network.
Maybe this is just a very quiet season for advertisers.
1000 impressions is not at all enough to monitor how good or bad the eCPM is. Looking at my admob stats, there are hours when I get 0 eCPM and hours when I have $6 or more eCPM. For 1000 impressions per hour. It also depends on the time as USA is 7-10 hours behind India for example. So if you are doing experiments while USA is sleeping then it’s normal to have a very low eCPM since Asia gives you the most impressions during that period.
I am using Playhaven and eCPM is around $1.5 for a game and $5 for another.
You are right about the statistical variation (for low data) - however, I was basing the revenue expected on the “stability” of the revenue results I was seeing for Admob (now) and Leadbolt (earlier).
However it is nearly 2500 impressions over the 2 apps - and you are correct that even with Admob and the earlier Leadbolt there WOULD be days where the revenue would have shortfall.
Your point about mistiming the test to not include sufficient U.S. traffic could be relevant also - since U.S. traffic accounts for bulk of revenue.
Your points about hourly variation are also significant - however those reflect “lack of statistical significance” of 1000-2000 impressions - but it could also (and more so) represent variations due to the time-zones waking up. So hourly variations itself will not indicate statistical significance of 1000 impressions.
A daily accounting of revenue WOULD however give better indication whether 1000 impressions are significant enough. I can confirm that there is considerable variation for 1000-2000 impressions also - but not if you repeat the test 2-3 times (as I have repeatedly tested Leadbolt recently).
A significant source of variation (in the daily revenue numbers) however COULD come from straddling of 24 hour period (for a day) - if it straddles the most busy U.S. traffic period then there COULD be significant variation (as just shifting start time of the period by one hour left or right could cause significant variation) - conversely if the 24 hour period straddles the “Asia is awake” time - then there would be considerably LESS variation (since the bulk of U.S. traffic would have occurred well away from the boundaries of your 24 hour period that the ad network is counting as a “day”).
However, your points about your own experience with Admob and also your performance with PlayHaven are relevant. This is heartening to hear.
Maybe I should tighten the belt and run a whole week test for:
Leadbolt
PlayHaven
AppFlood
Note how each developer individually is testing each of these ad networks - what a waste of time/resources !! Much easier to do for big publishing houses or if you have 10 apps (or high volume) - could allocate 10% of traffic to test performance of the ad networks on an ongoing basis.
Well, ads from Admob are totally different to the ones from Playhaven. It depends a lot on the type of app you have.
I am having around $0.7 eCPM with Admob and around $4 with Playhaven but revenues from Admob are a lot better because the banners are on the screen all the time while Playhaven shows interstitials between levels.
Each app is different so you should probably try different ad networks in your apps and get some good statistics before you decide which one to choose and focus on.
In my case I had just switched the interstitials form Admob to Leadbolt. The banner traffic still was going to Admob.
So I was documenting the change in revenue for Admob vs. the gain from Leadbolt.
UPDATE: I now have an update on the figures I gave above - as I’ve seen recently at least, Admob seems to REVISE the revenue figures it shows for a certain day. So even though it has started to show a bar for “Today” (suggesting the accounting for yesterday is complete and revenue for yesterday would be final) - that is not the case. And Admob revenue for yesterday can see some changes later.
As has happened now - oddly the revenue for Admob has recovered the $7 loss I complained so bitterly about above (maybe Google being the all-seeing entity actually listens … sigh).
So I am seeing zero revenue for today (which corroborates the observations in this thread which complains that Admob showing zero revenue for them today):
However, the revenue for yesterday has been boosted back to nearly usual daily levels - so I haven’t lost as much revenue from Admob as I had stated a bit earlier (because Admob has now revised it’s revenue figures) - and thus the slight gain from Leadbolt is just extra - there could be a number of reasons for this (maybe revenue was generally higher yesterday anyway - higher impressions/users etc.) or just advertisers better performing yesterday/better ads etc.
In any case it throws the spanner in the works for the “outrage” argument. Though it still does not address that approx. 2500 interstitial impressions were thrown Leadbolt’s way - and delivered $0.25 in revenue for that day.