About the "match ratio"

The match ratio describes the similarity between the users on a first-party audience list to the users on a third-party audience list. Numbers greater than 1 are a positive correlation, where higher numbers indicate a greater similarity. Match ratios under 1 correspond to a negative correlation, and generally imply a list isn't good to target.

Match ratios, like other audience-related metrics in the Audience Composition report, are calculated using a sample of all applicable bid requests from the past 30 days.

How to interpret a list's match ratio

Generally, higher match ratios denote a greater similarity. However, when evaluating a list's match ratio, please consider the size of the list. If a list is small and has a high match ratio, it may not have enough scale to be worthwhile as compared to a larger list with a high match ratio.

Formula for calculating match ratio

                                     Cookies on both first- and third-party lists
   "Probability of Overlap"            ――――――――――――――――――――――――――――――――――――――――
                                       Cookies on the third-party audience list

  
      … divided by――――――――――――――――――――――――――――――――――――――――――――――――――――

  

                                      Cookies on the first-party audience list
        "Baseline"                    ――――――――――――――――――――――――――――――――――――――――
                                                All possible cookies


  
 
                                                  = "Match Ratio"

In other words:

  • The Probability of Overlap is the dividend, which is the odds of a third-party audience list cookie being a first-party audience list cookie as well.

  • Probability of Overlap is divided by the Baseline, which is the odds of any cookie across all inventory sources a partner has the ability to buy from being in the first-party audience list. 

  • This equation produces the Match Ratio, which describes how much more likely it is for a member of the third-party audience list to be a member of the first-party audience list than it is for a cookie chosen at random (from all cookies across all inventory sources) to be a member of the first-party audience list.

Was this helpful?
How can we improve it?