Currently we are experiencing a very high sync time for our LinkedIn ad analytics connector for which we have raised the support tickets a few times and also got some suggestions from LinkedIn team. Some of these suggestions were implemented as part of the support tickets however there are few suggestions from LinkedIn team for which we were suggested to raise a feature request. This high sync time is hampering our marketing teams decisions/optimisations process as the campaign performance gets reported after almost 2 days. Kindly refer the details below and let us know if you think this is something that can be picked on priority.

Potential improvements with a reduction in extracted data

An example sync of the connector shows over 20GB of extraction vs less than 500MB of loaded data. This indicates there could potentially be large reductions in the data the connector is extracting from the source.

Dashboard:

 

 

 

Potential improvements recommended by LinkedIn

Recommended suggestions from LinkedIn:

1. Reduce API calls by reducing aggregation
2. Pull less data, ensuring only needed data is pulled
3. Parrelize the API calls for different tables

 

Info from LinkedIn discussion:

• One recommendation is that you don't aggregate data (if possible). For example, if you need a monthly report, then just pull a monthly report, don't pull daily report and add up the metrics for monthly report yourself. This way it'll use less API calls.
  
  • If I understand correctly, this is the first improvement which you have mentioned and is already implemented
• Can you optimze the schema pull less data? if some metrics are not needed for analytics then don't include them in the API, it save the data transfer volume. Can you optimize the sync frequency making it less frequent as it's needed?
  
  • This is something already optimized from your end as you have mentioned in additional improvements. We’ll do another round of check on our side to see if we can further reduce the number of tables we are syncing. However, I am not sure whether there is a way to only select the required field within a table, as we realised that there are multiple fields within tables which we sync but never use
• About the frequency of making API call to LinkedIn, it's something specific to the partner implementation, e.g. to shorten the E2E time -> making parallel API calls, making less frequent API calls etc. We suggest you consult with Fivetran about this optimization of LinkedIn connector. If Fivetran supports parallalize the API call in one LinkedIn connector already, then you may not need to use multiple LinkedIn connectors.
  
  • Can you please check on this point and let us know your views/suggestions on this