DAU over-reporting, along with client version: "UNKNOWN'

Context:
Data captured using
"com.unity.services.analytics": "4.4.1",

We recently soft launched our app on IOS and Android in Canada a couple weeks ago, and we are attempting to validate the metrics and analytics being captured.

We noticed that the counts of DAU and MAU appear too high, and don't match direct data from Apple and Google.

When exploring via Data Explorer, in our production env, when I look at a particular day (9/3/23) I get the following:

Using Metric DAU grouped by Platform
DAU - Anrdoid: 91
DAU - IOS: 79

Using Metric DAU grouped by version
DAU - 1.0: 54
DAU - 1.0.1: 40
DAU - UNKNOWN: 109

Using Metric DAU grouped by 'None'
DAU - 170

Question 1 is: How can the DAU for a particular day, when I only change grouping, give me different TOTALS of players? (platform yielded 170, version yielded 203,)?!?!

Whats worse, the UNKNOWN values are very high and account for the highest number of DAU.

When I try to dig into that oddity of version "UNKNOWN". To see where that is coming from
I have run SQL Data Explorer queries to see whats going on.. and I can see for one particular example that is odd:, 3 sequential events with the same USER_ID where the CLIENT_VERSION is inconsistent:

USER_ID, EVENT_NAME, CLIENT_VERSION, EVENT_ID, EVENT_TIMESTAMP

88522e79a20864f298c91923fc0aa64f
clientDevice
1.0.1
3164778771312160770
2023-09-03 13:53:26.138

88522e79a20864f298c91923fc0aa64f
newPlayer
1.0.1
3164778771299577856
2023-09-03 13:53:26.130

88522e79a20864f298c91923fc0aa64f
gameStarted
UNKNOWN
3164778771307966465
2023-09-03 13:53:26.130

Note how the "gameStarted" and "newPlayer" events, both had the same user ID, but the client version is different? And they both appear to come in at very close to the same time.

Question 2 is: What is going on with the UNKNOWN values here?

(for more context, I will show the event JSON completely below as well for the gameStarted and newPlayer events)

Game Started:

{
  "buildGUUID": "99fa54ae08d1a41659f716ea67cd4029",
  "clientVersion": "UNKNOWN",
  "collectInsertedTimestamp": "2023-09-03 14:58:07.393",
  "eventDate": "2023-09-03 00:00:00.000",
  "eventID": 3164778771307966465,
  "eventLevel": 0,
  "eventName": "gameStarted",
  "eventTimestamp": "2023-09-03 13:53:26.130",
  "eventUUID": "9ed1bf46a9a11ef0d2fe0f3f91f225a5",
  "eventVersion": 1,
  "gaUserAcquisitionChannel": "None",
  "gaUserAgeGroup": "UNKNOWN",
  "gaUserCountry": "CA",
  "gaUserGender": "UNKNOWN",
  "gaUserStartDate": "2023-08-18 00:00:00.000",
  "mainEventID": 3164778771307966465,
  "msSinceLastEvent": 0,
  "platform": "IOS",
  "projectID": *REDACTED*,
  "sdkMethod": "UA1_SDKLESS",
  "sessionID": "2428195151612471000",
  "unityInstallationID": *REDACTED, but the same in both events*,
  "userCountry": "CA",
  "userID": "88522e79a20864f298c91923fc0aa64f",
  "userLocale": "locale"
}

NewPlayer

{
  "buildGUUID": "99fa54ae08d1a41659f716ea67cd4029",
  "clientVersion": "1.0.1",
  "collectInsertedTimestamp": "2023-09-03 14:58:07.393",
  "eventDate": "2023-09-03 00:00:00.000",
  "eventID": 3164778771299577856,
  "eventLevel": 0,
  "eventName": "newPlayer",
  "eventTimestamp": "2023-09-03 13:53:26.130",
  "eventUUID": "ad4b054ad5e6ac9155c9fbb4c579ef5a",
  "eventVersion": 1,
  "gaUserAcquisitionChannel": "None",
  "gaUserAgeGroup": "UNKNOWN",
  "gaUserCountry": "CA",
  "gaUserGender": "UNKNOWN",
  "gaUserStartDate": "2023-08-18 00:00:00.000",
  "mainEventID": 3164778771299577856,
  "msSinceLastEvent": null,
  "platform": "IOS",
  "projectID": *OMITTED*,
  "sdkMethod": "UA1_SDKLESS",
  "sessionID": "2428195151612471000",
  "unityInstallationID": *REDACTED, but the same in both events*,
  "userCountry": "CA",
  "userID": "88522e79a20864f298c91923fc0aa64f"
}

bump

Hi @dppc ,

Thanks for reaching out! I'll list your 2 questions below and do my best to answer them.

Question 1 is: How can the DAU for a particular day, when I only change grouping, give me different TOTALS of
players? (platform yielded 170, version yielded 203,)?!?!

DAU can appear to be higher if you group by categories that could have the same users in them. So if a bunch of your distinct userIDs sent in events with both versions 1.0 AND 1.0.1, they will be counted twice.

Question 2 is: What is going on with the UNKNOWN values here?

What is looks like to me is that you have a lot of UA1_SDKLESS events being sent in, which come from Legacy Analytics. I would guess that most, if not all of your clientVersion UNKNOWN events come from the sdkMethod UA1_SDKLESS.

I hope this helps!