The paradox of counter-productive advertising

I spent some money in advertising, in the very first days of life of my app.

It was money spent badly, because the added downloads to my app have not been enough to climb the “most recent free apps” list in that category.
As soon as the ad campaign finished, due to the lack of visibility, the downloads suddenly decreased.

Now I pay this as a lesson.

But the paradox is another one and I want everyone to think about it when planning an ad campaign.

According to latest Google I/O, not only installs are important, but also uninstalls, install duration and the global install trend.
The graph of my downloads makes a huge step downwards from one day to the other.
Due to the fact that daily downloads passed from x to x/10 in one day, I have:
the global install trend dropped → this is negative according to Google in the parameters for giving the app visibility
the uninstall rate is > than the install rate → this is because of the large differential in number between ad-inducted downloads and normal downloads.
the install duration is lower, because new installs are fewer than older installs.

I don’t have any data supporting this theory, but my suppositions are that, without the initial ad campaign, probably NOW the app would have less total downloads, but more daily downloads.

Think about huge campaign if you are not sure that they will lead to results!

This is a very interesting topic - and I have thought about this in passing as well - whenever I have considered “what to do” if one were to use ad-campaign (like appbrain) to push an app.

However, since I have not actually pushed an app, I have not devoted much time to examining it …

But your instincts seem right - if you artificially drive downloads with a huge peak (i.e. spend all in one day) - you are going to see a FALL in new downloads the next day (compared to today).

So certainly the “trend” will not be helping you.

Immediately this suggests that if you want to spend money - it should be spent in increments - i.e. small and then watch downloads. That is, spend it carefully.

I think AppBrain allows that type of spending pattern - allocating a certain amount per day.

You also point out that the uninstall rate experience is very important - and you point out a scenario where downloads from today (high number from ad-campgaign downloads etc.) may start to uninstall tomorrow and that have a huge uninstall number (greater than the installs for that day).

While this may be true - it is only so if a considerable number of today’s (artificially high) downloads decide to uninstall much later i.e. after a few hours or after 24 hours etc.

But my own sense is that the uninstall number (for most apps) will depend more so on the install number from that day. That is, for apps which are new (few recurring users) - you will find that most of the new users will uninstalls that same day or immediately after testing your app. I am just guessing here - but I suspect the likelihood of uninstall is probably HIGHEST right after install and first run - and then it probably falls down over time. In fact if a user has not uninstalled the app, it is likely to be forgotten to uninstall - until they sit down over the weekend to clean up the phone (or need more space for other apps - more of a problem for low-spec/priced phone demographic).

So if you have high new downloads today - your will ALSO have a high uninstall rate TODAY. Tomorrow will be tomorrows new downloaders uninstalling - plus a small proportion of today’s ones (who uninstalled a few HOURS later instead of immediately after first run) and then returning users - some uninstall rate from that.

For OLD apps - the returning user demographic WILL be significant - so you have 3000 new downloads in a day - but you have 2500 uninstalls - of which maybe 1500 are from new downloads and 1000 from your “Active User” base (which maybe high - something like 50K users - of which you WILL get a TRICKLE of users uninstalling).

After all that discussion - I suspect the “path of good sense” would be to SCALE into an ad-campaign install campaign. That is, to not force huge downloads the first day - but try to give it an exponential rise day after day - that SHOULD be picked up by Google as perhaps an organic growth path. However, you will not be able to keep that up for long - with a modest budget - so then the second choice is to watch the new downloads coming in - and try to add maybe 10% to each day’s download numbers.

Be aware though that with ANY type of installs - i.e. search-based or ad-campaign-originating - each download will lead to uninstalls as well (whether those uninstalls fall in the same day or the next day or whatever as discussed above).

There is a danger that with ad-campaign-based installs that there maybe a HIGHER uninstall rate - though it may not be surprising if ad-based ones have BETTER retention than user-originated-search-based installs - I mean that one can conjecture - but one has to look at the data and there maybe anomalies which contradict “common sense” - at which point one will examine what/why and then conclude that “oh - that is because of THAT” etc.

But generally it has been suggested that “incentivized downloads” i.e. those from Tapjoy/GetJar etc. (where user gets SOMETHING in return for downloading an app) - that the uninstall rate will be higher. And it seems that non-incentivized installs therefore cost more also (i.e. AppBrain vs. Tapjoy/GetJar). The reason given is that the user is in a hurry to rack up some coins and will download crap apps as well - only to uninstall them immediately i.e. see the app download as a hurdle/means to get the coins fast and now …

So you may find uninstall rate different between:

  • downloads of your app that come from user searches for keywords
  • downloads from ad-campaign (either admob banner or the better results-yielding AppWalls i.e. AppBrain and others)
  • downloads from incentivized (Tapjoy/GetJar etc.)

CONCLUSION
Maybe try a campaign where you spend an increasing number over time - starting off small the first day - i.e. match your spending to the downloads.

So if second day get 50 downloads - then for third day try to spend to get 25 downloads (50% increase) - then next day when you get say 50 + 25 - then you buy 50% of that for the next day.

This will make the graph look like exponential to Google - the problem is how long you can continue that - and how much can you save using THIS scheme compared to some other spending scheme - and this is EXPONENTIAL - so it will swamp any other scheme anyway - so will wind up paying hugely this way also.

But if you decide that you are going to do it for 10 days it may not be that bad:

day 1 - spend nothing - watch how many downloads get
day 2 - spend nothing - watch how many downloads get - say 50
day 3 - spend to get 50% more downloads than prev day - say get 50 + 25 = 75
day 4 - spend for 50% of 75 = 37 - say get 50 + 25 + 37 = 112
day 4 - spend for 50% of 112 = 56 - say get 50 + 25 + 37 + 56 = 168

If your app is really good, it will get a top position.
pushing the downloads just results in a high uninstall rate and google is NOT stupid, they will recognize your tactic

Right - making an app that is liked by users and viral also probably helps :slight_smile:

Don’t forget that Google lives on ads, so probably it’s not completely true that good apps will get to top position anyway.
For example, the app category are carefully hidden to users in Google Play. Users rarely search for a specific category.
This means that no matter how good your app is, to make the real jump you have to appear in the “global” statistics (and by the way become featured), and you can do that only by spending A LOT in ads.
So, the one-day boost might be negative, but Google might also have the tools to detect it, as soon as they really don’t want to penalize developers that pay for ads.