Traackr’s Relevance Science

Over the weekend, we released improvements to our Relevance scoring. This improvement isn’t a change in the way we compute Relevance, but rather the way we express it. As a result, you may notice your A-lists looking a bit different. By improving the way we normalize Relevance on a scale of 1-100, you can now get a better expression of a person’s true expertise at a glance, even before taking a deeper look into people’s profiles.

We have always expressed Relevance on a scale from 0 to 100, and this has not changed. However, the way we interpret those values has. Originally, the influencer with the most posts and keyword matches on the list would be assigned 100 Relevance and everyone else on the list was scored relative to that person. So, someone talking on the topic half the amount of that person would have a Relevance of 50. While this may have been straightforward, there were cases where it was not convenient. There was no simple way to interpret the “quality” of an influencer by just reading their Relevance score. On one list, 80 Relevance could mean “moderately relevant expert”, and on another list it could mean “highly relevant expert”. To figure that out, you would need to check the posts and compare with the #1 influencer.

Our other two metrics, Reach and Resonance, are global across all the influencers within our database. After having analyzed thousands of A-lists, we have figured out how to adjust our Relevance computation algorithm to express our Relevance score in a more global way, similar to Reach and Resonance. With our new formula, 90 Relevance means the same thing everywhere. Anyone above 50 would be true experts. Close to 100 means that they are not only topical experts, but they publish a lot online. And close to zero, they have mentioned the topic, but not enough to be considered experts.

Just to clarify a few things with the new formula:
  • There may be no one on your A-list with a perfect 100 Relevance. This is because with the new formula, someone has to publish very frequently and have a significant amount of relevant posts. So someone with 100 Relevance wouldn’t only be an expert in the space, but an extremely active expert as well.
  • You may find that there are more people with lower Relevance than you are used to seeing. This is not a bad thing, it is just representing their expertise in the space much more accurately. You have to remember that our rankings are driven by all three metrics, so it is very possible that someone can have very low Relevance, but high enough Reach and Resonance that they still make the list.
  • Low Relevance may not mean true expert, and will likely be volatile. If a person has a lower Relevance, say around 10, this means they have mentioned the topic enough to be considered influential, but may not necessarily be an expert. These people will most likely be among the most volatile on the list (moving on and off the list frequently).

We hope you will like the new Relevance and find it more intuitive and actionable. You may want to check the ‘scores’ tab in the Analytics Suite where influencers are divided in four quadrants. These are now easier than ever to interpret because they allow you to cluster people in groups. For example, anyone in the ‘high Reach+Resonance / high Relevance’ quadrant would be considered “experts with a large audience”.