AdColony Performance Issues on Amazon Since March Plugin Update

I did a build on March 27 and my AdColony inits on Amazon devices dropped by 66% the day after I published the update.

I didn’t change anything relating to advertising and my initializations on AdColony from Google Play users continues to match my daily active users, which leads me to the conclusion that the plugin update has had a negative impact on Amazon devices.

Has anyone else noticed a sharp down-tick in AdColony performance for Amazon devices since this plugin update?

I noticed on the AdColony page in the Marketplace that the plugin’s last update is now listed as April 3, 2019, so I did a new build of my app using Daily Build 2019.3474. When I checked it out on ADB, plugin.adcolony is listed as 2.1.3( SDK 3.3.5) which is lower than what Rob announced on March 4 (SDK 3.3.8).

I’m getting a successful initialization of AdColony on my device, but I don’t know if that’s because the problem is fixed, or if it’s because my device is one of the 33% of devices that are still working with AdColony post update.

Any thoughts on this?

The checkin commit comments say 3.3.8. Let me check with engineering and verify.

Rob

Thank you, Rob.

I published an Amazon update on Friday with the new build mentioned above. According to analytics, on Saturday 90% of active devices were using the update, however there was no change in inits on AdColony. I can say with a fair degree of confidence that the April 3 update to the AdColony plugin has not solved my problem.

Any insight into this problem would be greatly appreciated.

This is very strange, because update was literally just dropping new AdColony library and rebuilding the plugin, without any code changes. No code changes were made.

I’m in the same boat- I didn’t make any code changes to the module that handles AdColony and I didn’t change my Amazon build.settings. My changes were limited to UI changes in scenes that don’t even “require” any of my advertising modules.

If there are any logs I can furnish or diagnostics I can run on my end, please let me know.

Is anyone else having this issue?

It’s distinctly possible that the new AdColony SDK has changed their behaviors and that’s effecting what you see. Have you considered asking AdColony directly if this is a change on their end?

Rob

Rob,

Yeah- I filed a support ticket last Wednesday. I’ll be reaching out directly tomorrow if I haven’t heard back.

Colin.

Rob,

I’m communicating with the folks at AdColony, and one of the things I noticed while sending them the Charles logs was that while ADB lists the plugin as “plugin.adcolony: 2.1.3 (SDK: 3.3.5)”, in the Charles logs it is “version”: “3.3.8”.

Anyway, AdColony is on the case, so hopefully we can get this sorted out.

Colin.

Hi Rob;

Colin asked in an earlier post, “Has anyone else noticed a sharp down-tick in AdColony performance for Amazon devices since this plugin update?”.

I haven’t noticed a “down-tick” as I have just begun adding AdColony interstitials to my games as an advertising option (in my waterfall). But, AdColony is showing some really wacky low DAU numbers and “Ad Requests” that are less than 10% of the actual AdColony inits and “load” calls made to their system. Very strange and poor performance. BTW – I know that the AdColony numbers are this low because other ad companies report accurate numbers 10 times higher.

The app update with AdColony was built with build 3468 in late February and the first day I “turned on” AdColony in the waterfall was March 20.

Best;

Steve

I would recommend that you put some prints in your adColony waterfall to print the event.response value and see if it’s a simple case of no fill available.

Rob

Hi Rob;

Thanks. Yup, I’ve done that.

My “init” call to AdColony is invariably successful. This is what is returned:

=================  AD EVENT - AdColony Listener  =================

Key: phase = init

Key: provider = adcolony

Key: name = adsRequest

Key: isError = false

=============================================================

My “load” calls are almost invariably met with silence. Though not always. Less than 10% of time (far less I am guessing), an ad gets loaded. The “silence” extends to the fact that the listener never receives a “loadFailed” phase response.

Note that after several days of my app (about 10K DAU on Amazon – which is where I am trying this initially) on thousands of devices calling “load”, the control panel over on AdColony will show (today’s example) 112 Requests and 112 Impressions – giving AdColony a glorious 100% Fill Rate! But in the same period my app has inited and called “load” many, many thousands of times.

When, in my waterfall, the opportunity for an AdColony interstitial comes up in the sequence, I do check for “adColony.isLoaded” and when that returns false, I do, of course, trigger a new “load” call by my app. Also usually met with silence (no “loadFailed”. 

And, of course, my test devices have excellent Internet connections and other company’s ads load.

This may all be unrelated to what Colin is describing. And my next step may be to contact AdColony support.

But it is both odd and disappointing and frustrating . . . and may be a problem on either side (Corona or AdColony).

Best;

Steve

Rob,

I heard back from AdColony today. Here’s the problem (quoting from the email I received):

“We had pushed a backend controller update at the time of [your sudden drop in inits]. The reason for the drop was that we started seeing a widespread issue where a NullPointerException was being thrown on Android OS 5.x and Amazon OS5.x devices; this affected apps using our 3.3.7 and 3.3.8 Android SDK versions and was fixed in our 3.3.9 Android SDK update. Do you have plans to update to a our latest SDK version soon; our latest Android SDK is v3.3.10 and can be downloaded from our GitHub at https://github.com/AdColony/AdColony-Android-SDK-3. Updating to our 3.3.10 SDK will allow ad serving back to these users.”

So it looks like, through no fault on the part of the Corona team, it was the updated AdColony plug in that caused my sudden drop-off in initializations. I know it’s a pain to have to update something that just got updated, but is there any chance of moving the AdColony plugin to SDK 3.3.10 anytime soon?

Thank you,

Colin.
 

I’ll see what we can do.

Rob

Hi Rob;

I’m monitoring this also as AdColony doesn’t work for my apps either.

I’m glad that Colin got the info from AdColony that they “broke” things on their end.

Thanks for looking into this.

Steve

Rob,

Thank you- I appreciate everyone’s time and attention to this problem.

Colin.

Ad colony should be latest 3.3.10 on Android and 3.3.7 on iOS now.

Thanks Rob and Vlads!

You guys are the best. We appreciate it very much.

Steve

I just pushed out an new version to Amazon with the updated plugin and I’m already seeing an uptick in inits on AdColony, so it’s looking like, “Problem solved.”

Thanks, guys!

The checkin commit comments say 3.3.8. Let me check with engineering and verify.

Rob

Thank you, Rob.

I published an Amazon update on Friday with the new build mentioned above. According to analytics, on Saturday 90% of active devices were using the update, however there was no change in inits on AdColony. I can say with a fair degree of confidence that the April 3 update to the AdColony plugin has not solved my problem.

Any insight into this problem would be greatly appreciated.