Facebook Ads Relevance Score Decoded

Back on February 11th, Facebook announced the beginning of the rollout for a “relevance score” with ads on their platform.

In case you missed it, here’s the article from their site.

I highly recommend you read it.

This has mostly been glossed over by others as I’ve seen virtually no chatter about it…

However, I believe this is the most important change to Facebook’s algorithm for 2015.

So, in my opinion, understanding it and how to manipulate the factors that affect relevance score will be the key to successfully advertising on Facebook this year.

Facebook says “Relevance score is calculated based on the positive and negative feedback we expect an ad to receive from its target audience.

The key in that statement is “we expect”…

What indicators does their algorithm look at in order to predict the positive and negative feedback they think an ad will receive?

It can’t be anything inherent in the ad itself, can it?

I’m writing this article on March 3rd of 2015 and I published an ad a few hours before to test this theory.

Take a look at the screen shot below…

Screen Shot 2015-03-03 at 10.45.54 AMAs you can see, this ad has been approved and has NO interaction yet and the relevance score is listed as dashes.

Because my agency manages six figures of ad spend per month on Facebook…

I have the unique benefit of looking at LOTS of data.

So I started looking into several ads from a few of our larger volume campaigns to try and figure out what influences the relevance score.

I decided to only look at ads within the first 10 days to make sure the data was fresh.

My first hypothesis is that click thru rate has the biggest impact on the relevance score.

Below are screen shots of the first three ads I looked at to try and prove this hypothesis.

Before you look at the data here’s some important things to know about this campaign:

1) These are three separate ad sets each targeting one specific interest group.  All three interests are different.

2) The campaign objective is “website conversions”.

3) The bidding strategy is oCPM and we’re optimizing on the conversion pixel.

4) The placement is “desktop all” meaning its running in the newsfeed and right hand column from the same ad.

Screen Shot 2015-03-03 at 10.31.14 AM Screen Shot 2015-03-03 at 10.31.26 AM Screen Shot 2015-03-03 at 10.30.59 AMIt would appear based on these ads that my initial hypothesis was correct…

As any good scientist knows, one sample is not enough to prove a hypothesis.

So I went to check another high volume account to verify my findings.

Here’s some things you should know about this campaign:

1) These are two separate ad sets each targeting one specific interest group.  Both interests are different from each other and each is different from the first three I showed above.

2) The campaign objective is “website conversions”.

3) The bidding strategy is oCPM and we’re bidding on “website clicks”.

4) The placement on these is desktop newsfeed only.

Screen Shot 2015-03-03 at 10.31.48 AM Screen Shot 2015-03-03 at 10.31.37 AMAs you can see, this sample blew my initial hypothesis out of the water…

An 11% CTR is just insane and that only resulted in a relevance score of 5.

A 2.2% CTR is also nothing to scoff at as well and this only gave a relevance score of 4.

Now this really made me have to step back for a moment.

I realized that comparing these two samples is also not accurate in testing this hypothesis…

There are too many variables that aren’t isolated to confirm the results.

1) We have different audiences

2) There’s different placements

3) There’s different bidding strategies

So here’s the next test I’m starting today to figure out if CTR really has a big impact on the relevance score.

I’m going to submit three ads in separate ad sets.

  • All three ad sets will be in the same campaign with “website conversions” as the objective.
  • All have the same copy, but a different image as image has the biggest influence over CTR.
  • All three ads will use the “desktop newsfeed only” placement.
  • All three ads will target the same audience and demographics.
  • All three ads will use the oCPM bidding strategy and bid for “website clicks”.

This way, the only variable in the equation is the image.

Everything else is exactly the same…

So we should get different CTR’s and be able to evaluate the relevance score as it directly relates to CTR.

I will be publishing the results of this test on my next blog post in a few days.

Stay tuned 🙂

UPDATE:  I published the initial findings of this test here.

As an FYI, this initial evaluation also brought up several more questions about what affects the relevance score.

Is it the bidding strategy?

Is it the goal of the campaign?

Is it the demographics?

I will be creating additional tests to isolate those variables and publishing the results too…

As you can see, I’m extremely thorough in tracking and testing the advertising that I do for myself and my clients.

That’s why we’ve been able to generate over $3.5 million in sales in the last 12 months exclusively using Facebook ads…

All on about $1.5 million in ad spend 🙂

(who else tells you how much they spent to make a certain amount of money LOL)

If you’re not getting at least a 200% ROI from your Facebook advertising, something is wrong…

But don’t worry!

I have a solution for you…

I’ve put together a free 11 point cheat sheet that reveals the steps you need to go through in order to make Facebook ads profitable.

These 11 simple and tested steps work in every market we’ve tested them (it’s been 17 and counting)…

If you’d like to grab a copy of this cheat sheet, just click the link below: