What is a “Likelihood Score”?
If you want, or expect, deep granularity and detail in your reporting, then knowing your likelihood score is extremely important. The Accuracy Reports are gold!Peter Bacon, Onboarding Manager
If you’ve never used CallTrackingMetrics, you may not be familiar with likelihood scoring, as it’s unique to our platform. The likelihood score basically states how confident we are that we matched your incoming call to the exact person who visited your website. Users can survey their likelihood score, and its changes over time, by producing an Accuracy Report.
The likelihood score refers to how confident our system is that a particular website visitor made a particular phone call. The score for each match can range from 1-100%, with 100% being the most confident.
Last week, Erika Rollins told us that Accuracy Reports were an incredibly important tool for her as a modern-day marketer. So I checked in with Peter Bacon, our Onboarding Manager, to learn how he explains this feature to the clients he works with every day.
Peter, how do you explain accurate call scoring to our users?
The way that I like to explain our likelihood score, is that it uses an algorithm which draws upon several factors to determine how certain we are that we tracked your call to the exact visitor to your website. I often joke that this algorithm is “voodoo” our engineering team came up with.
Basically, we’re matching clicks to calls. When some people hear the term “likelihood score” or “Accuracy Report”, they’re concerned that it refers to the accuracy of our call routing. Of course, we’re always going to route your incoming calls wherever they need to go, that’s never a problem. The likelihood score is just referring to matching those incoming calls to whichever online ad that caller clicked, or whichever path they took to get to your website. The closer that score is to 100, the more confident we are that we matched your calls to the specific session data of your online visitors. Of course, that means the likelihood score is only relevant to callers who visited your organization online.
When you your likelihood score hits a low percentile range, we may not always be able to connect your calls to the proper session ID. That’s why we recommend setting up many different tracking numbers: in order to ensure the greatest possible accuracy in your call tracking. This is especially important for users that experience very high call volume in their organization… matching clicks to calls in that scenario may get skewed without enough tracking numbers.
How exactly is the likelihood score determined?
Our algorithm draws upon a variety of factors to determine which website visitors are most likely to have made each of your phone calls, such as time, geography, activity, and history. In order to keep your likelihood scores as close to 100% as possible, we recommend at least one tracking number for every twenty daily visitors to your website from a particular source.
Really, in order to keep your likelihood score in a healthy range, just make sure that you have enough tracking numbers. If too many visitors – from many different sources – are all seeing the same tracking number every day on your website, then you may see your likelihood scores falling away from 100%. We will always try match a visitor to your calls, but if you want those matches to be as accurate as possible, I recommend sticking to that guideline I mentioned before – at least one tracking number for every twenty visitors.
Any thoughts on call tracking without knowing your likelihood score?
That would really depend on what someone is trying to get out of their call tracking efforts. If you want, or expect, deep granularity and detail in your reporting, then knowing your likelihood score is extremely important. The Accuracy Reports are gold!
If you’re a CallTrackingMetrics user, you can either look in your call log for the likelihood score of each incoming call, or produce an Accuracy Report by navigating to “Reporting”, then “Accuracy Report”.