Google Analytics connector missing startdate/enddate parameters

Hi Thomas,

Microsoft did indicate in this tread that they were considering/planning to ad the start/end date parameters. Perhaps someone from Microsoft can chime in. I ended up proxying my requests through a middle-tier service to get what I wanted.

 - Chris

Yes, that is true. But sampling isn't an issue when querying one or two months, as my reports will do, but for longer timeframes (the start-date, end-date parameters you'll be adding to the cube) the sampling increases. My reported issue is, I assume you now agree, that you were doing client-side aggregation.

If you really want to get clever with your connector, then you have to have the query folding logic send multiple queries back to GA to reduce sampling. Google documents that approach in their blog, etc.

For example, if I query ga:yearMonth with start-date = jan-2014 and end-date = dec-2015, then for a busy non-premium GA site I'll likely get numbers that are slightly or way off. But if I split this into 24 separate queries with start-dates and end-dates bracketing single months, then I'll avoid sampling errors - at the price of making more queries.

Hi Tristan!

Any news on this, my problem besides propper aggregation, is that I get sampled data instead of actual values, and since I'm reporting sales my data is not real.

If I use a date range filter, the original query (confirmed with fiddler) is for the whole range of the DB (going all the way to 2013) and then the filter is applied, so i only get the results from the sampled filter... so no, its not useful.

Any news on this? is it going to be fixed/addressed soon?

The only current way to use the GA connector is to allow it to query back to the beginning of time and then to filter. So sad.

Yes it's sad, because its useless in a lot of scenarios, for you is Users for me is sales (i just cant report them out of sampled data). I guess implementing this shouldnt be that hard!

Hold on. Sampled data is a different matter - not an issue with the connector semantics. If you want unsampled data, you need GA Premium.

not really, if the query is small enough (lets say less than a year) GA won't sample data (in their API) premium or not! If you catch the request with fiddler you can see that PBI is requesting for sampled data! if we could run smaller queries like you ask, i would have what i need. (BTW tableau handles this part just fine... just that im on trial)

Can you share the Fidder request that shows it requesting sampled data?

Or are you saying that because there is no timeframe specified that PBI is requesting the entire timeframe and that is implicitly causing sampling.

so there it is, but i guess this is the response, so as you can see we recieve sample data, since the start date is really far off. The same report using google api but with a date range of 1 year, and i get unsampled data! 

Ok, so you are getting sampling because of the implicit date range. So I'll extend my "so sad" to cover that also, as there is currently no way to avoid it.

Hi,

It's a long shot but has anyone found a solution to this without using a third party service as an middle man?

I agree with the many other posts without being able to pass a date filter it is forcing sampling and redering the connector mostly useless. A number of posts across the Power BI forums confuse the issue and suggest post filtering.

So far my many attempts to work round the issue have failed. :(

Thanks

Thomas

Hi intrasight

Thanks for the post, somehow I started to have the same issue just a few days ago. Do you have any news about the fix or the workaround? 

I have the same issue just with dates and E-Commerce data. Only that if I want the correct data as my Analytics UI I have to filter by weeks. :( 
I don't know why the folding is so bad for the data for some properties and not for others. 

Let me know if there is an Idea or a Bug issue to vote for. 

Best,
J.

I've not heard any updates on this. I assume it just not a high-priority item :(

Hey, Chris!

I have also encountered the data sampling in Power BI.

It even occurred within small projects because Power BI required data for the whole period.

I have found the solution in creating our own connector that would divide queries by days and import them in the SQL Server Database. This way, we managed to solve the problem.

Eventually, our solution escalated into a separate product. It is more convenient to work with one database which contains not only Google Analytics data, but other data as well. Such as data from Google AdWords and CRM systems.

You can try to set up integration on our website.

I am always there and happy to help if you have any questions.