Marketers and agencies need to stop thinking about look-alikes in their audience and focus on act-alike, says Ben Willee.
In the offline world, talk of truly personalised marketing has pushed marketers to wean themselves off the teat of demographic segmentation and latch onto psychographic targeting. Yet in digital, marketers are going gaga for look-alike targeting. You’d think we had already learned that this approach can be as hit and miss as a standalone Helix persona.
Still, platforms such as Facebook continue to promote look-alike modelling to advertisers as a way to boost ad relevance to an audience. But these targeting methods, for the most part, rely only on demographic information or things the user has told us about themselves. The truth is, you can’t trust what people say they do. You have to see them doing it, you have to say ‘show me, don’t tell me that you like Brand X’.
Just because you are a fan of the Facebook page doesn’t mean you’ve made an actual purchase from the business.
Let me give you an example. Let’s say I’m trying to target in-market car parts buyers using a look-alike model. I take my existing customer pool as my seed audience segment. These people would probably have the following characteristics:
- male skew,
- age 25-to-49,
- DIY interest,
- don’t play online games,
- follow football, and
- slightly more likely to use a mobile than a desktop or laptop.
Creating a look-alike based on this seed audience would help me find more people who match the base profile. The problem is that people who look-alike don’t necessarily act-alike. This is where most advertisers, and their agencies, are at today. This is also where we have a problem.
Right now, most advertisers are targeting people with the same demographics and showing them the same ad (or a variant of it) every time we see them online. And they say the definition is insanity is doing the same thing over and over again and expecting different results.
The act-alike alternative
With act-alike audiences, marketers don’t have to rely on what consumers say they do. The technology can follow them and actually see what they’re doing. As the name suggests, these audiences are based on actions, not characteristics.
So what triggers tell me that a person is in the market for my products, and what value can I give to those signals? Let’s take a person who is the complete opposite of the previous example:
- 16 years old,
- no previous interest shown in cars or car parts,
- no DIY interests,
- plays online games,
- no interest in football, and
- much more likely to use a mobile than a desktop or laptop.
She may not look like someone who would be interested in purchasing my product but by analysing her actions, including brand website visits or searches for auto accessories we can clearly see she is exhibiting all the trigger characteristics of someone interested in purchasing my products.
By using an act-alike model she would be programmatically shown my ad campaign and in this case, she is likely to convert to a sale because she is purchasing a birthday present for her father.
While marketers may be slow to adopt it, act-alike isn’t some newfangled phenomenon. Tech platforms have been using these audiences for some time explaining act-alike as supersets of look-alike modelling. These supersets rely on moments of influence, aka opportunities to show an ad, that are similar to previous moments which have worked, rather than focussing on users that might look the same.
Such moments are identified by a combination of demographic, behavioural and contextual data.
It’s time for advertisers to beat their addiction to look-alike audiences because it’s actually doing them more harm than good.