Analytics 2.0 Connector in Power BI - Has anyone used it successfully?

[mattd37159866]

We are trying to use the native connector in PBI to report on our Adobe Analytics data.  Unfortunately, any time we try and do anything remotely close to the type of reporting we need, it does not work.  Either will take for-ev-er to even try and load the rows or will just error out. 

I've investigated with Adobe & MS and it seems it is the way the 2.0 API works that causes any worthwhile reporting (more than 1 dimension & 1 metric over a daily time period) hits the limits of the 2.0 API calls. https://developer.adobe.com/analytics-apis/docs/2.0/guides/faq/

Even trying something simple like: 

- Daily data for a 6 month time period

- Marketing Channel 

     - broken down by Device Type 

- Visits

- Orders

- Revenue 

this should be a "simple" report, yet it hits the API limits and effectively renders PBI useless for us.  

Has anyone use PBI w/AA connector successfully?  

[fhusain · Accepted solution]

Hey @mattd37159866 The issue here could be due to the time period. I agree this is a simple request, but the way API works is bit complex.

To illustrate, consider fetching the top 10 Marketing Channel line items, which initially require just 1 API request. Now, breaking each line item down with Device Type will make it 1*10 requests. Then you add daily granularity, so 5 columns which will be multiplied by 6 months of data with daily granularity. This can be too much to request from an API request, and what if you try to reduce the time period to 30 days? Does it work?

You can also try fetching monthly data and then combine the data into one dataset. Fetching all the data at once may lead to query failure.

But, if you feel this is still an issue, please feel free to log a support ticket via Admin Console to us. We would need the Power BI trace logs which should have Adobe API Request/Response.

[md2420]

Thank you @fhusain .  What you describe is exactly the conclusion that we came to after going round and round with support tickets for both Adobe & Microsoft.  It is the API complexity that is causing too many requests to come in at once, even with "simple" requests.  The answer from Adobe was "working as designed", which I guess it technically is.  Except the design is such that it does not allow for any meaningful use of PBI as a reporting mechanism for Adobe Analytics data (which is very disappointing). The API should be configured differently to not multiple these calls like you describe (and we discovered).  30 days of data does us no good and I can't see setting up a different/new connection for each month of data I want (as well as any further detailed information across dimensions that we'd want to report on).   

All of this to say, I am still curious to know if ANYONE is actively using the AA connector in PBI and have success with it for real business uses and reporting.  I have yet to hear (or find) anyone that does.  

[fhusain]

Yeah, I can understand. So, one of the client actually used to pull monthly data and then collate into a single dataset and yes, it is something not straight forward. As the connector is built on the API, it does have some limitation.

But what I can suggest instead of using connector you can take dumps from the Data Warehouse and import the data directly in the Power BI. I'm not sure how much feasible it would be at your end by this something you can try.

[md2420]

Thank you for the quick reply and further thoughts. Yes, we've considered the DW option especially now that we can send to Azure and other containers instead of just email or FTP.  The issue I have with that is as the data changes (like filters/segments), the live connector is better served as those changes happen instead of static outputs from Datawarehouse.   

As an example, we have a "bot traffic" segment that we are constantly updating as we find and suppress bot traffic that we identify separately from the OOTB suppression that AA has.   So that would change metrics/counts for past data as we add more conditions to that segment.