Search Unity

Sudden drop in number of daily installs on Google Play Store

Discussion in 'General Discussion' started by AxPetre, Jun 23, 2018.

Thread Status:
Not open for further replies.
  1. red_unit

    red_unit

    Joined:
    May 25, 2017
    Posts:
    5
    Nope. Seems like apps on Android native are suffering too.
     
  2. pep_dj

    pep_dj

    Joined:
    Nov 7, 2014
    Posts:
    154
    Yes, I can confirm. Mine is Android native, and dropped too.
     
  3. gencsadiku

    gencsadiku

    Joined:
    Feb 4, 2017
    Posts:
    5
    I hope there will be a solution soon by Google, cause this situation is so depressing :(
     
  4. shiggsy_dev

    shiggsy_dev

    Joined:
    Jul 12, 2018
    Posts:
    6
    Happened upon this thread and thought I would chip in my experience, as there seems to have been a small step in the right direction lately.

    I read on the Android Blog that Google had changed the discoverability algorithms' so I went and had a look. The apps I am focused on are in the Finance Category, there are 4 apps which provide the same service, an official app and three un-official apps, a main keyword search shows these 4 apps all together, and, in the past, they all appeared in each others Similar Apps list. After the blog post though, the 'Similar Apps' shown now were all completely un-related, not in the same category, it looked more like a random selection.

    I sent a non-response feedback from the Dev console, saying the 'Similar Apps' was broken, with screen shots, this was back in May. I kept an eye on it and this week I have noticed a slight improvement, one of the apps is now appearing in each of the other apps similar list, and the official app is displaying apps, though not really similar, they are at least all from the Finance Category.

    One thing I noticed in this thread from one of support replies was this :

    "To ensure the Similar Apps section promotes discoverability and shows new or different apps to users, the apps are algorithmically selected and change regularly."

    The key phrase being 'different apps', which seems a weird thing to put seeing as 'different' is the opposite of 'similar', but I am assuming that different in this context means a different set of similar apps, and not just a just a set of completely different apps. You can imagine that this is the bit where they have got too clever and broke it.

    You'd think a search engine company would have done better..
     
    Joe-Censored and gencsadiku like this.
  5. cthulhu2duo

    cthulhu2duo

    Joined:
    Apr 13, 2015
    Posts:
    3
    When i saw drop in my game installs, i thought it was some fluctuation or some console analytics error. Now i'm desperate.

    I left my job a month before this drop to make games in my home-made studio. Depressed :(
     
    noumanriaz and gencsadiku like this.
  6. samulew

    samulew

    Joined:
    May 6, 2018
    Posts:
    3
    I know how you feel. After two years of working on nights and weekends I did the same. Left a job in May. All was going well until June 21st. A little bit depressed :)
     
    gencsadiku likes this.
  7. cthulhu2duo

    cthulhu2duo

    Joined:
    Apr 13, 2015
    Posts:
    3
    I hope Google will resolve this problem in our favor. It's interesting, that company, where i worked before (i still have access to their stats) has 50-150% boost in app installs started around June 21-22.
     
    gencsadiku likes this.
  8. rigidbuddy

    rigidbuddy

    Joined:
    Feb 25, 2014
    Posts:
    34
    A friend of mine told me that same story happened to Youtube since May. Niche (indie) channels just do not discover any more via search and side blocks. And the situation didn't change since then.

    Also my ex-colleagues from one big company told me that organic installs on their project suddenly increased since the end of June.

    Maybe Google ranks bigger projects high ranking algorithms just prefer bigger projects from now on? Kinda store monopolization
     
  9. samulew

    samulew

    Joined:
    May 6, 2018
    Posts:
    3
    What I believe is that they are slowly deleting bad apps while the new algoritm is learning. Until learning process is finished, they prioritize apps that make them the most money (or those they feel that make them most money). But I am also pretty sure that they did not plan it exactly like this and that they messed pretty bad but will never admit
     
    cthulhu2duo likes this.
  10. gobitgames

    gobitgames

    Joined:
    Jan 15, 2016
    Posts:
    11
    We don't have a GDPR consent dialog in our ad supported games yet and, in fact, we took that games off the European app stores at the end of May until we finish new versions with GDPR consent support. The big download drop, however, hit us a month after we took the games off the European stores.

    One of our games used to appear in a collection and although I can still find the collection on the web version of the Google Play store, the collection doesn't appear in the Android version of the store on any of my devices. I think that could easily be a big reason for our drop.

    I'm also pretty sure that when I looked at the google search cached version of some similar games page, we used to appear on them, but now we don't. Unfortunately, the google webpage cache is now too new to confirm this theory.

    We write "native" Android apps programmed mostly in C++ using my modified version of PopCap's open source SexyAppFramework - I think we may be the only people in the world still using this engine! Though, we do use Unity Ads as one of our ad providers.

    As for guessing how the new algorithm works, we have 5 games in the Google Play store that we can use as data points (does anybody have an App Annie premium subscription that they could use to grab way more data?):
    • Burger Shop (Play first 20 levels for free, pay to unlock the rest)
    • Burger Shop FREE (Play the whole game for free with ads, pay to remove ads)
    • Burger Shop 2 (Play the whole game for free with ads, pay to remove ads)
    • Burger Shop 2 Deluxe (Premium - Buy the game)
    • Evil Orbs (Play the whole game for free with ads, pay to remove ads)
    And here's some data on the downloads, ratings, User Acquisition Reports, and Android Vitals reports.

    data_points.png

    The user acquisition data comes from the retained installers report in the Google Play console Acquisition report using the month of May and using the Measure by Country Play Store Organic and looking at the all countries and US data. The numbers in parentheses compare the performance to games in the same category using the same monetization model - sort of a nifty report that I didn't know about before.

    In terms of the Android Vitals, I'm of the opinion that games with ads have more crashes and ANRs because the ad libraries themselves have more of these problems.

    The easiest thing for me to see from the data is that Burger Shop 2 suffered the worst dropoff in installs, but if I compare it to Burger Shop FREE (the other Burger Shop game with ads), it clearly has a better rating, mostly has better retention rates, a worse ANR rate, and comparable crash rates... though none of these are anywhere near bad behavior.

    And if I compare Burger Shop FREE to Burger Shop (ads vs. try-before-you-buy), I see that Burger Shop FREE clearly has a better rating, and better retention rates, but much worse ANR and crash rates than Burger Shop, yet Burger Shop suffered a slightly worse dropoff in downloads.

    My cynical side wonders if the change has to do with revenue per download. We don't make much money per download since we mostly only serve ads and allow people to pay to remove the ads. Could the key algorithm change involve a low revenue penalty? Also, it may be that even Google doesn't know what's going on since they say they're using machine learning algorithms which are famously hard to audit. Maybe all Google knows is that they've slightly raised some overall metrics that they wanted to raise...

    So what have I figured out here? Uh... nothing? All I know is that I like Evil Orbs but not many other people do. Sad!
     
    AxPetre and Avietry like this.
  11. Stan22z

    Stan22z

    Joined:
    Jul 15, 2013
    Posts:
    22
    why you can be so sure? So you believe that after sometime will back to normal? because similar section is so messed up. At the moment it looks to me that they are erasing indies part by part. because this happened already on 2017 and nothing happened. We just didn't feel it. Until this update.
     
  12. Stan22z

    Stan22z

    Joined:
    Jul 15, 2013
    Posts:
    22
    https://gadgets.ndtv.com/apps/news/...-install-rates-google-play-algorithms-1876405

    "Having said that, quality apps are the need of the hour for Google Play as several issues emerged in the past that involved malicious apps. Last year, Google Play leveraged machine learning efforts to remove more than 700,000 "bad" apps and 100,000 developers, a 70 percent jump from the preceding year. "In fact, 99 percent of apps with abusive contents were identified and rejected before anyone could install them," Google had claimed in a blog post."
     
  13. pnyxtr

    pnyxtr

    Joined:
    Oct 29, 2014
    Posts:
    10
    Well my game was top 100 grossing in the US, in its category, before this happened so I don't revenue is the key here.
     
  14. gobitgames

    gobitgames

    Joined:
    Jan 15, 2016
    Posts:
    11
    Thanks for that data point.

    How was the revenue per install on your game, though? For instance, was its download ranking higher or lower than its grossing ranking?

    I'm wondering if Google might be prioritizing higher revenue per download.
     
  15. Stan22z

    Stan22z

    Joined:
    Jul 15, 2013
    Posts:
    22
    Anyone from US here who has got huge drop installs too?
     
  16. gencsadiku

    gencsadiku

    Joined:
    Feb 4, 2017
    Posts:
    5
    Mine dropped from 5k daily to below 100 a day
     
  17. christovstoi9

    christovstoi9

    Joined:
    Jun 30, 2018
    Posts:
    5
    is there anyone who has dropped the installations first and then increased to the same level before the sudden fall?
     
  18. Prometheusheir

    Prometheusheir

    Joined:
    Aug 15, 2015
    Posts:
    19
    13EAB66A-30D2-4AD9-B6B8-22233A3B5E1E.jpeg Two of mine are now practically dead on Google Play. And doing fine on Apple and Amazon. So Google is now officially too risky to count on in terms of income source.

    EDIT: btw nothing wrong with crashes, retention or ratings from what I can see. However it shares 90% of functionality with an older app of mine so maybe a ‘clone’ flag was triggered.
     
    Meltdown likes this.
  19. shiggsy_dev

    shiggsy_dev

    Joined:
    Jul 12, 2018
    Posts:
    6
    So instead of just providing feedback, I decided to raise a publishing ticket, I couldn't detail the issue completely with the character limit, and waited for the support response so I could provide examples of the issue. I got back the same automated \ paste response that others have had, "This behaviour is normal and to be expected..." blah blah.

    So I replied with some example apps and what they were showing in their Similar Apps Section, pointing out that these are very specific Finance based apps, whilst the apps being displayed as 'Similar', came from the Categories of : Entertainment, Communications, Video Players, Lifestyle, Personalization, Tools and Medical !, they were not at all similar and quite clearly it isn't working.

    To his credit, this time the support person took a look at my examples and agreed it wasn't working as intended, documented the issue and has escalated it to the technical team, which I know is what others have had as well.

    Have to wait and see now I guess.
     
  20. tiburon

    tiburon

    Joined:
    Oct 10, 2013
    Posts:
    19
    Thank you Shiggsy, I think everyone of us should do the same thing. I’ll do it.
     
  21. pnyxtr

    pnyxtr

    Joined:
    Oct 29, 2014
    Posts:
    10
    Download ranking at best 280 in the US
     
  22. Ogien

    Ogien

    Joined:
    Nov 21, 2012
    Posts:
    156
    Glad I found this thread and I can see I am not the only one in this boat.

    My downloads have also dropped significantly (-40%) on Google Play. My app is still #1 in the store for my target keywords so it seems to be related to the similar apps category.

    The reason for the change is the Google modified their app search algorithm to account for retention of users. I am guessing some very addictive games with expanded in in-app stores will do much better now. Those happen to be large games with big teams (in many cases)

    Honestly I have been expecting something like this for a while. Things could not continue like this forever. Historically no market can be dominated by Indy type companies/individuals forever. We had a good run but this may be the start of a real indypocalypse.

    It is sad, I don’t think it will be good for the customers, there will be less creativity, less experimentation.

    My advice is save your earnings while you have them, focus down on the 1 game that makes you money and make it more awesome focusing on customer retention. Merge with some other teams if you can to get bigger. Start making Clash Clans clones?
     
    Last edited: Jul 15, 2018
    Meltdown likes this.
  23. neoshaman

    neoshaman

    Joined:
    Feb 11, 2011
    Posts:
    4,209
    You assume the indypocalypse would one big event, but it's actually a lot of smaller one, saturation of digital store front, rise of quality and shift in platform policy are all contributing to bury profit down.
     
    xVergilx and Meltdown like this.
  24. Meltdown

    Meltdown

    Joined:
    Oct 13, 2010
    Posts:
    5,420
    Haha... enough of those out there already...
     
  25. Ogien

    Ogien

    Joined:
    Nov 21, 2012
    Posts:
    156
    I said start
     
  26. neoshaman

    neoshaman

    Joined:
    Feb 11, 2011
    Posts:
    4,209
    It has already :D it's propagating :eek:
     
  27. MalyaWka

    MalyaWka

    Joined:
    Jan 2, 2015
    Posts:
    11
    Everybody salute!

    I do not remember if I wrote that the problem with "Similar Apps" actually did not appear on June 22, but much earlier. It is somehow related to the user's credentials. Back in May, the "Similar Apps" section worked oddly - if you looked at it in incognito mode (without logging in to the account) it worked correctly, and if you looked at it after login, everything was bad.

    June 22 all became bad - and logged in and not logged into Google Play.

    Now they are apparently still trying to fix something, as normal "Similar Apps" appeared on the application page again if they did not login to Google Play. But if you open "Similar Apps" section, then everything is still bad there, and there are no applications that were on the main page.

    Sorry for my Promt English.

    Screenshots for understanding:
     

    Attached Files:

    cthulhu2duo and PZ2100 like this.
  28. Stan22z

    Stan22z

    Joined:
    Jul 15, 2013
    Posts:
    22
    nice. did you report it to GG?
     
  29. MalyaWka

    MalyaWka

    Joined:
    Jan 2, 2015
    Posts:
    11
    Yes. Waiting for an answer.
     
  30. gobitgames

    gobitgames

    Joined:
    Jan 15, 2016
    Posts:
    11
    Thanks again for the info! So, like you were saying, that would be evidence against Google biasing search results based on revenue per download.

    I'm going to assume for now that the Google Play recommendation algorithm went beyond its original programming, simulated all nearby universes, and arrived at the conclusion that suggesting Jewels Classic rather than Burger Shop would create the best of all possible worlds :)
     
    Last edited: Jul 16, 2018
  31. LilGames

    LilGames

    Joined:
    Mar 30, 2015
    Posts:
    189
    I think now that everyone has it wrong, Similar Apps algorithm is not what is broken; it's the store that has not been updated yet to have the heading "You Might Like:" instead.

    ¯\_(ツ)_/¯
     
    Last edited: Jul 17, 2018
  32. PZ2100

    PZ2100

    Joined:
    May 27, 2017
    Posts:
    19
    Ok, I got a phone call from a Google person (their Play Store promotion team). I asked about the drop of installs. His replies:
    • They are aware of the drop in installs
    • Everything is how Google intends it to be (it's not a bug)
    • From now on, it will be like this. If they decide to change it, there will be official announcements
    • We should start promoting our apps/games with ads. If our LTV is too low to cover UAC, we must work on improving LTV. Since LTV varies across countries, it may be easier to match LTC and UAC in some hand-picked countries and scale up at least there.
    • Almost no one at Google knows how apps in Similar Apps are picked because it's governed by machine learning. Not even the principles or factors included in the decision are known. Or, at least, that's secret even to Google's employees not directly involved, and especially the general public. Google is ok with Similar Apps lists as they are now
    • Despite the secrecy, he could confirm that LTV is a major factor considered by the AI in ranking/suggestions
    My thoughts/hopes:
    • Since it's machine learning, there is still a chance that, after some time seeing decreased store views -> installs conversions on Similar Apps and similar suggestion tools, this AI (machine learning) could decide to get to a state closer to that we had before and we would start seeing some of our downloads come back. But this restoration will be at best slow and gradual, and at worst won't even happen. Obviously, it depends on how fast the AI learns, how much data it needs to make decisions, and if actually your apps are more profitable to Google (directly or indirectly) than Talking Cat Tom et al
    • One should actually work on increasing LTV (if your game genre/type allows it)
    • Since 95% of app traffic is users who will never ever (like, never, never, never, never ever) pay... - the AI should recognize that at some point and make a conclusion that in general it's pointless to re-direct this traffic into apps with high LTV. Instead this traffic should at least be monetized on ads of something outside the store (like, physical goods or financial services), not of other apps or games. I'd then expect this traffic (actually 95% of global traffic) be directed back to our apps. Even to an extent that we'd see more traffic than before the change. However, one also needs to keep in mind that games with high LTV (like Clash of Clans) still need a certain ratio of non-paying users simply because their economies and mechanics would collapse if every user were paying for boosts. These games also need swathes of players (even of low quality) to make themselves look popular, so that the paying players felt cool and stayed. I guess machine learning would find a fine balance between how much of this 95% worthless traffic it would keep in the high-LTV games (to provide good enough environment for the other 5% paying users) and how much it could re-direct to ad-supported games. Ultimately, this boils down to some non-obvious ratios between ad revenue from the 95% non-paying and in-app revenue from the 5% paying. Machine learning is exactly there to tackle this fuzzy and complex math task (of course, with the goal of maximizing Google's revenue).
    In any case, the gold rush is over.
     
    Last edited: Jul 17, 2018
  33. yourstoryinteractive

    yourstoryinteractive

    Joined:
    Mar 27, 2017
    Posts:
    46
    Well it reads believable, until you meet a developer like me whose worst (judging in terms of LTV) app quadrupled and the best one (25x LTV) decreased 10 times in terms of downloads. Let's hope that the infant AI will eventually grow to become smarter.
     
    tiburon likes this.
  34. Stan22z

    Stan22z

    Joined:
    Jul 15, 2013
    Posts:
    22
    any answers from them yet?
     
  35. gencsadiku

    gencsadiku

    Joined:
    Feb 4, 2017
    Posts:
    5
    Guys algo will take some time to restore things in their places as before, stay positive and keep working
     
  36. pr3t3nd3r

    pr3t3nd3r

    Joined:
    Sep 26, 2015
    Posts:
    11
    Ni
    Nice, i said it from the start ... when google breaks stuff with bugs they will say it's a feature and not only not fix it but make it worst. The gold rush time is over way before this, many of us where just picking up the leftovers ...which they decided that they also want/need ...

    Another question, with the assumption that the traffic no longer comes from related suggestion and only from search, the quality of users should increase. But my impression is that also the new installs are worst quality ... poorer retention. (lool ... even the stats for retention are not present for July .... just bug over bug ...sure it's intended, I'm sick of asking them to show me the stats real time not with days delay ... and now it seems if month ... wtf ... until i have access to the stats the game is already penalized by the algorithm if something goes wrong with a release is not like indies have a huge testing team to assure each release quality and lack of bugs ... )
     
  37. Meltdown

    Meltdown

    Joined:
    Oct 13, 2010
    Posts:
    5,420
    Yay, so the rich get richer.. :(
     
  38. ekstro

    ekstro

    Joined:
    Jul 18, 2018
    Posts:
    4
    Google, if it's not a bug why we have empty categories on all our accounts (Google Play Web)? Also what about differences in "Similar Apps": on my account I don't have any but in incognito everything is ok.
     

    Attached Files:

    Last edited: Jul 18, 2018
  39. gobitgames

    gobitgames

    Joined:
    Jan 15, 2016
    Posts:
    11
    We aren't currently waiting for any answers from Google Play support. The email I quoted above was their answer to us when we asked if they knew whether reduced installs were being caused by search/recommendation bugs or the new algorithm.

    Since he said that they knew about the search issue and were working on a solution, I have some small amount of hope that things could improve at some point. Though, I do have experience with Google knowing about another bug and seemingly not being able to fix it for over half a year.

    Also, to anybody who has the Google Play promotion team tell you that the current behavior is correct and exactly as expected, I think we have some reason to doubt this because:

    1. Other Google teams have acknowledged that there's a problem.

    2. The promotion team is trying to sell ads so they may claim that everything is working as intended and that the only way to succeed now is to buy advertising.
     
  40. LilGames

    LilGames

    Joined:
    Mar 30, 2015
    Posts:
    189
  41. LilGames

    LilGames

    Joined:
    Mar 30, 2015
    Posts:
    189
    Just logged into my dashboard and noticed the total lifetime installs number is gone. (You know, the column that shows Current/Total, now it only shows Current).
     
    gencsadiku likes this.
  42. gencsadiku

    gencsadiku

    Joined:
    Feb 4, 2017
    Posts:
    5
    same here
     
  43. einEntwickler

    einEntwickler

    Joined:
    Jun 26, 2018
    Posts:
    13
    Today Google was ordered by the European Commission to pay €4.34 billion ($5 billion) for unfairly pushing its apps on smartphone users and thwarting competitors. Google has been ordered to stop the practices within 90 days, or face additional penalties.
    https://money.cnn.com/2018/07/18/technology/google-eu-fine-android/index.html

    It could be that issues in the PlayStore are currently not the highest priority...
     
  44. Murgilod

    Murgilod

    Joined:
    Nov 12, 2013
    Posts:
    5,821
    I'm 9000% willing to bet that the people involved with the Play Store algorithm change and the people involved in this don't even carpool together.
     
  45. Avietry

    Avietry

    Joined:
    Mar 4, 2015
    Posts:
    21
    This is unrelated to Play Store IMO. Google Search/Chrome has always been an issue in Europe.
     
  46. bart_the_13th

    bart_the_13th

    Joined:
    Jan 16, 2012
    Posts:
    440
    yep... I've noticed it too...I think its about 3 or 4 days ago they made that change.. maybe they want to emphasize us not to focus on the number of total downloads but instead focus more on maintaining active users...
     
  47. PZ2100

    PZ2100

    Joined:
    May 27, 2017
    Posts:
    19
    You are absolutely right.. The gold rush was over already in 2011, when everyone understood that every game must be free. If you'd ask anyone at the time what an IAP should be, they'd say "more levels?" Of course we all found out later that selling content (and much else) just doesn't work.

    Yeah that's weird, and probably means that Similar Apps suggestions used to be more accurate than search results.
    From my own experience as a user, searching for a concept/genre would only return maybe 1/3 of relevant results (to me). But if I just take my favorite game and then get sucked into the black hole of chain Similar Apps suggestions, I would gradually discover the whole sub-space of games I really like. Most likely (as others here have suggested) the Similar Apps list is now filled with temporary stubs which people at Google have hastily hand-picked from the top-charts.
     
    Last edited: Jul 20, 2018
Thread Status:
Not open for further replies.