The Right Audience, The Right Time

Since iOS14.5 hit the scene, audience testing has become much less fun. With less data to optimize with, once stellar audiences like lookalikes have fallen by the wayside in favour of more broad, open-ended, algorithm-reliant audiences. If you asked most marketers what their top performing audience was these days, the bulk of them would say ‘broad audiences’, or audiences that are left open-ended with only some small variation of age, gender, and location segmentation.

That said, not all hope is lost - it is still possible to have fun with audience testing while still being efficient with your spend. But how? The answer lies in segmenting your broad audience using interests. Not only is there a large library of interests to choose from for segmenting, but the signals that Meta bases these interests on come from on-platform behaviour. For example, if you ‘like’ a page about dogs, Meta knows you are interested in dogs. ‘Interest’ audiences, for this reason, don’t need to rely on dwindling pixel data to conclude, as they can retain all data within their own network.

To further illustrate this and to prove that this isn’t just a matter of opinion, I conducted a survey recently asking people which of the big three audience types they are finding the most success with, removing broad from the equation. The result showed that a staggering 60% said that interest audiences, in some form, are outperforming lookalikes.

Interests Are In

So, interest audiences are performing well for people, that’s great, but how do we use them effectively? Well, there are two main ways to use interests in your targeting:

• Adding a single interest to your ad set or the more popular option:
• Adding a ‘stack’ of interests to your ad set

While adding a single interest to your ad set can be used to more broadly segment your ad set targeting, using a ‘stack’ is a much more popular approach. A ‘stack’ simply layers a series of interests together in a single ad set, which will more dramatically affect your segmentation.

Within interest stacking, you are presented with two options: AND or OR. Each time you add another interest to your ad set, your audience effectively becomes either smaller and more refined or larger and less refined, depending on which modifying conjunction you use. If you use AND, your audience becomes smaller, and if you use OR, your audience becomes larger.

How To Hack Interests

To supercharge this, though, we can dive into a thought experiment. 

Using the AND logic in the audience segmentation tool, we can quite granularly hone in on a subset of an audience that we can uniquely sell to. But how do we find the right interests to target? Simply ‘thinking’ of unique audiences to target is difficult. 

Let’s pretend we are selling running shoes for women. The obvious interests to target here are ‘running’, ‘marathons,’ ‘athletes,’ etc. What isn’t immediately obvious is something like ‘dogs.’ This is just an example, but let’s explore how we landed there.

To do this, we will use something that I am calling the Interest Venn Diagram.

The concept is relatively simple but should yield some wonderful results. Essentially, you take two different, relatively obvious interests within your audience and begin brainstorming on the type of people that exist at the intersection of the two. Of course, using two fairly similar interests, like running and athletes, isn’t going to yield great results, so instead, I recommend pulling in some Google Analytics data here.

So, let’s say we want to use ‘running’ as one interest for this test. That makes sense - we are selling running shoes, which is a fundamental interest for this category. Great, so what do we layer over that? Let’s pretend we are looking at our Google Analytics account for this brand. We will open up our affinity category and in-market segment breakdowns to find information about our website’s visitor behaviour. In this example, let’s pretend that ‘Pet Lovers’ is a top result. From this, we can think, ‘runners are highly active - what type of pet might they have?’ from which you might deduce that it’s more likely they have a dog that they can play with, as opposed to a cat which is usually more home-bound and passive. From this, you now have a new interest to target: dogs.

Sure, you can let the algorithm automatically find these people if you combine ‘running’ and ‘pet’ in an ad set and choose AND as your modifying conjunction, however experimenting yourself allows you to ‘stack’ a series of these Venn Diagram interests together in a single ad set using OR as your modifying conjunction. One of the main reasons that the OR modifying conjunction traditionally works so well is that it gives the algorithm more data to play with. With a larger potential audience to target, it can be more efficient in finding cheaper conversions first. To do what we’ve done inside an ad set automatically, we need to limit the audience by using the AND logic. In contrast, now we can stack a series of Venn Diagram interests together, providing Meta more flexibility to find you the cheapest results possible. 

Obviously, the above example is somewhat straightforward, so it’s important to spend some time brainstorming on many different variations. I personally like the idea of laying this out in a Venn Diagram to clarify better what characteristics you are trying to combine. This part of the process is both tedious and fun - let your creativity run free here. There are no wrong answers.

Using audience characteristic data from your audience insight tools, such as Google Analytics, is very important here. Let’s say, for argument's sake, there are only two types of runners, which are very different: those who work corporate jobs and workout to influence their social status and those who are in tune with their mind, body, and soul who workout to nurture this sense of self. Now, what if your audience insights tools say that your average purchaser is mostly interested in ‘Home & Garden’? If we had drawn our own conclusions, we might have targeted those from the first group and potentially burned money where the real audience was the second group. Using this example, we can use our Venn Diagram to add in new, previously unseen adjacent interests such as ‘horoscopes’, ‘environmentalism’, etc. Using this exercise, with all steps working in harmony, can be quite revealing about your audience, which can help you develop unique targeting strategies.

Baby Steps

Step 1: Take your core audience interests‍

Step 2: Find adjacent interests from audience insight tools‍

Step 3: Brainstorm interests that arrive at the middle of those two‍

Step 4: Use these results either individually in an ad set, or stacked in a single ad set using the AND modifying conjunction.

Step 5: Have fun

Step 6: Profit?