KNOWLEDGE BASE

Understanding Google Analytics Sampling in Tableau Deskop


Published: 09 Dec 2016
Last Modified Date: 12 Dec 2016

Question

What is "sampling" in Google Analytics and how does it affect Tableau Desktop?

Environment

  • Tableau Desktop
  • Google Analytics

Answer

If the view is based on sampled data, the numbers in Tableau will differ from those in Google Analytics. However, both Google Analytics interface and Tableau will alert the user that data is sampled.

What is sampling?

  • Google Analytics back-end stores raw session data.
  • Back-end also stores standard reports, which are updated on a daily basis from the raw data. 
    • Standard reports are not sampled as data is incremented every day. An example of standard report: Sessions by country (Audience > Geo > Location).
  • When a non-standard report is created (e.g. a custom report is created in GA web portal, or an extra dimension is added to a standard report - either in GA interface or via API call), Google Analytics needs to query the raw data.
  • If there is too much raw data to process, it uses sampling (i.e. processes only a small percentage of the data and fills in the gaps by extrapolation)
​Note: If more than 500k sessions (25M sessions for Premium Analytics or Analytics 360) need to be processed, sampling is used.

Why do results differ when sampling is used?

Each time sampling is used, a different (random) portion of the raw data is sampled, so results will be different.

How to avoid sampling and make sure all data is returned?

Avoid non-standard queries and only request dimensions/metrics included in Standard reports.

    Additional Information

    For more information, see How Sampling Works in the Google Support pages.
    Did this article resolve the issue?