Previous month:
November 2016
Next month:
January 2017

The Relevancy Of Relevancy Scores And The Lie Of UI

Your computer is lying to you. Or rather, the user interface is giving you information that you mistakenly believe is accurate because it Looks Official

Some examples:

LinkedIn Zipcodes: Why do you think they are accurate? Are they inputted and checked? Has the person moved since starting a LinkedIn account? Is the distance from the zipcode the distance from their house or from the city center of the zipcode? 

We don't know. I bet the developer never thought about it - because developers don't think about such things. 

LinkedIn employee count: LinkedIn filters people by the size of their company. While they can add Fortune 1000, how do they track the number of employees?  It's almost assuredly by the number of employees on LinkedIn, which is why Construction companies have 1000 employees on LinkedIn and 5,000 employees overall.

Facebook Search: The number of Female Developers who work for Citibank in Boston. Do that search. That should suffice as an explanation. 

It's not just the searches. It's our systems.

What's the Relevancy Score in Facebook Ads? I asked someone yesterday, but then asked them to tell me what the score meant without using the word "relevant."  It's harder than you think. Especially since none of us "really" know, including Facebook employees who are just repeating what they are told.

Relevancy Score has a meaning to Facebook, of course. It's the number of times people click on your ad.  Based on that score, Facebook determines if you're a good or bad advertiser. which affects how often your ads show up. They want the better performing ads to show up more.

Stop and think about that for a second. Relevancy Score is how often you spend money with Facebook. That's useful to Facebook. It's useful to Advertising firms, who are paid on the percentage of their spend. Is it useful to the rest of us? Is it relevant to our goals using Facebook ads? Maybe.

And yet - try this - go ask someone in social what the relevancy score is. They'll say it's a score that determines how relevant your ads are. Relevant to what? Relevant to who? They'll struggle, and then probably throw something at you.

Again, it's not their fault. It's a kind of short-circuit that occurs. Relevancy Score, 10 miles from Zipcode - these seem like they are carefully calibrated to be accurate data. They are not. They are tricks (some honest, some not) of the User Interface, much like the 36,284,312 results on a Google page.

One of the challenges we have in recruiting social people is the parrot effect, where jobseekers tell us what they've heard, without thinking it through. The best hires - don't just repeat the words. They are not fooled by a Title Field.