Imagine the followings:
You have a big bowl full of balls of many colors. the bowl represents a specific country (say the USA) and the balls are the unique users of that country.
We take from the bowl only users of one color (users which browsed on a specific month a specific site, say homedepot.com).
After taking this "homedepot" balls we ask what is the probability that among these balls we have users of another site (say, menards.com).
The probability (generalized in few ways) is the affinity score you see on the site.
Since besides the size of the bowl, all other parameters are not symmetric (the unique user of the site or the conditional probability) we wouldn't expect same values.
SimilarWeb Audience Interests allows you to see the list of websites with a high probability of being visited by the same users as the analyzed site.
The Affinity score in ‘Audience Interests’ is the measurement of how frequently users visit two websites, relative to the average distribution of visits within the analyzed country. A 50% affinity denotes a “null affinity” - in other words, the overlap between two sites is the same as the average across the entire population. Any affinity score above 50% represents a greater overlap than the average, while an affinity score below 50% represents a lesser overlap than the average.
Using this metric
Use this metric to discover where your audience is spending time to broaden your targeting lists; and understand which other solutions/ offerings users are evaluating or comparing your offering to.
For more information how to best utilize this feature, view this guide: Getting Started with Audience Interests