When calculating data for a tag analysis, we leverage probabilistic counting – a technique by which data can be compressed and analyzed to return real-time insights.
Probabilistic counting is used only when your audience includes a Tag or a Tag in combination with Resonate survey attributes, 1st party data, or 3rd party data.
Prior technology created an artificial limit on the number of attributes we could have in the platform – limiting our ability to add new attributes, first and third party data attributes. With this limit lifted we can:
- Accept 1st party data at scale via JS tag or batch load
- Onboard 3rd party data providers more easily (eg. Location data, retail data, etc)
- Perform more complex UI interactions without degrading speed (eg. Unique composition across target audiences)
A small margin of error for data processing is expected. The margin of error increases with increasing audience complexity, so the more tags that are added to an audience definition, the greater the margin of error may be. That same probabilistic counting referenced above is used for tag metrics which is why hits and uniques differ.
Please sign in to leave a comment.