Google Analytics returning incorrect data

[DjMillward]

Hi All

We've noticed that we have a mismatch in data being returned from Google Analytics through the Power BI Google Analytics connector and what can be seen on the GA Dashboard. 

Let's take for example the first metirc which is User for the last Calendar month.  This should return 181,582 according to the Dashboard. 

In Power BI I do the following :

> Get Data > 'Google Analytics' 

> Drill down to the 'Website Data - ALL' Folder .

> Select 'Users' from the 'User' folder

> Select 'Date' from the 'Time' folder 

> Click on Load . 

On a fresh page I can now use a card visualisation to display the Users total .  And a filter to limit the date - by the last calendar Month  ( which I've  checked the start and end date using DAX ) 

This gives me 213, 235 . Its a big difference ! .

Looking into this I found this post from February last year. 

https://community.powerbi.com/t5/Desktop/Data-from-Google-Analytics-Doesn-t-Match/m-p/136095#M58342 


Which states 

The API we're using for Google Analytics is the one available to free users, and that API is known to return less accurate values for high volume sites due to the way it samples data

As far I know we are not a Google Premium or Google 360  user.

Also I've tested the an API call ( v4 ) straight to Googles API and this gives me a result of 180, 985 . Which is close enough to just be an issue with timezones or something ! 

Any help with the Power Bi Google Analytics connection would be appreciated. 

[DjMillward]

In my case I have found the reason ( but not the solution ! ) 

I decided to bypass the connector and retireve the data straight from the Google API.  When retrieving the amount to month as a whole it was correct and then as soon as I split by date the total becomes incorrect. 

It's the same on the Google Analytics dashboard itself, get the result for the single days; and the total will different when you search for the days together. 

Take a look at this post https://www.en.advertisercommunity.com/t5/Google-Analytics-Reports/Number-of-total-Users-differ-over...

Here's an example : 

1st Jan -  Jon and David go to website . ( 2 users)
2nd Jan - David and Dan go to the website ( 2 Users )

But when we retrieve the result for the two days together Google returns 3 .  As Davids second visit is not included. 

So could I work this out in my calculations ?  

If I could get the returning percentage for a comparison of 'day 2' to 'day 1' that would be a possibility.  For example that would be 50% ; and I'd take it off the day total.  

As far as I know this isn't data that's available to me.   At present then it looks as if I can not use the time slicer on the Google Users stats if I want it to match those number seen on the Google dashboard. 

[v-cherch-msft]

Hi @DjMillward

You may have a look at below posts to check if they are helpful.

https://community.powerbi.com/t5/Desktop/google-analytics/td-p/77804

https://community.powerbi.com/t5/Desktop/Google-Analytics-Pulling-Incorrect-Data-Date-Hierarchy-Not/...

http://martin.boros.sk/powerbi-and-google-analytics-incorrect-data-one-possible-solution/

Regards,

Cherie

[DjMillward]

Hi Cherie


Thank you for your reply. 

Things I’ve tried 

1 . Updating PowerBi version to Septembers 2018 ( from April 2018 )  . 
2. Reduced and tested a query to just the very basic.  One column with 3 months of date records. 
3. Followed the blog

http://martin.boros.sk/powerbi-and-google-analytics-incorrect-data-one-possible-solution/  ; which tells me it is not a Sampling issue.  


None of the above helped and at present it looks like I'll need to use a solution other than the Connector