Why you shouldn’t over-rely on basic demographic targeting options

Why you shouldn’t over-rely on basic demographic targeting options

Remember the “good old days” of advertising? When ad campaigns had vision and really spoke to their target audience?

Me neither. If you actually watch Mad Men, it’s not a world any of us should want to emulate. Plus I’m too young to have worked in an age where the basic demographics of your target audience (eg. age, gender, race) were all you had to go on, and I bet you’re too young for this too. Why then would we make these factors the cornerstone of our campaigns? And yet this is something we see day in and day out.

It’s certainly true that when you are creating your business or marketing plan you should be developing target personas, as well as updating them as the direction of your business evolves (hello 2020 & beyond????). But it’s usually best to think of your persona as a diving board not the destination. It helps you make your marketing messages (website, ads, emails) get specific rather than generic. But this does not mean that it’s a good idea to just apply the core demographics from your persona to ad targeting.

Let’s say you are selling a veggie subscription box. You have research showing that the people most likely to get this type of subscription are professional women in their 30s with families (note: this is made up, we have no idea if this is true). So far so good, you might develop a core persona based on this hypothetical person. But then how to implement this in, say Facebook? When we look at accounts people have built themselves we’d often see the audience defined as something “women in their 30s“, often with nothing else! This highlights why over-focussing on demographics might not be good.

What’s wrong with over-focussing on demographic targeting?

  1. The above example focuses on age and gender as if that’s the most salient point about a person. But unless you think everyone in that demographic would convert (a remnant of Don Draper era targeting), all ad networks have much richer signals. In Google, someone might be searching for a veggie subscription box. In Facebook it might be available as an interest. Any of those are orders of magnitude more valuable than demographics but when you were developing your personas you probably wouldn’t have put “wants a veggie subscription box” in the description — that would be implied. So translating demographics to targeting too literally makes us more likely to stop looking for more valuable signals.
  2. If like us you’re in Australia, you will probably also be experiencing limitations in advertising due to Australia’s low population compared to say USA, UK or other countries. Audience sizes being too small is something we see for Australian advertising time and time again. To go back to the Facebook example, let’s say you add an interest for veggie subscription boxes but also keep the “women in their 30s” target. What you’re telling Facebook is essentially: “of all the people interested in veggie subscription boxes, narrow this down further to women in their 30s; for anyone else don’t show the ad”. Is this really what we want to be doing?
  3. Over-relying on demographics is often short-sighted in terms of your business development. It’s true that different ages, genders, races etc will convert at different rates. But if we think of our products as being primarily “for” a certain combination of the above we are much more likely to start replicating discrimination in the way our products and services are set up. Even if you remove demographics, your ad might still be served in a discriminatory manner, as this Vox video shows. Do we need to add to this ourselves?
  4. Finally for some verticals, targeting by specific demographics may be against the ad platform’s terms of service and/or illegal. For example you cannot target by gender on Facebook in posting ads for jobs. Chances are your industry is not on this smaller list, yet. But maybe it will be, maybe it should be!

So how would we recommend you deal with demographics? Let’s continue with the veggie subscription box business and pretend it’s a new website and ad account.

What should you do about demographic targeting?

  1. Unless your product/service offering really is restricted by age/gender/race (I bet it’s not), do not add any demographic targets from the outset but make them about the better targeting options offered by the ad platforms. In this case we might use Facebook interests, lookalike audiences, search keywords for Google and Bing, remarketing lists from your own website, custom affinity audiences in the Google Display Network and so on.
  2. The main exception to this might be if your target market is so large that you really do want to restrict your audience as much as possible first so as not to burn through your cash (again that probably means you’re not advertising in Australia or have an absolute mass-market product).
  3. Before turning on the ads make sure to turn your Google Analytics demographic reports on, as well as making any relevant updates to your privacy policy/user privacy controls.
  4. Once the ads are running for a little bit and you have some conversion data, check your Google Analytics demographic reports to see how different values for age and gender convert compared to each other. GA does not have race as a field, we’re not fully in Black Mirror yet. Note that these reports will not include anyone for whom age + gender is unknown which may be 50% or more of your traffic. Here’s an example of how to easily see this:
  5. If there’s a substantial difference, in Google and Microsoft ads, add bid modifiers for these demographics. Use the proportional ROI difference; for example if 18-24s convert at 75% less than the baseline, make their bid adjustment -75%. This would mean you don’t stop ads for any demographic, although the bid might be greatly reduced.
  6. For Facebook ads this option doesn’t exist at the moment so you would need to (A) segment your existing ad sets by demographic or (B) update your audience definition to narrow the audience. Option A has the added advantage of being able to tailor the ad copy to the demographic (also a possibility in Google/Microsoft ads although it can be a lot of work).

Following the above steps does not guarantee that your ad campaigns will not be served in a discriminatory way. Nor does it guarantee best results. It is however how we recommend most advertisers deal with striking an appropriate balance.

Got a question?

For something quick, the best place to ask would be on Twitter or in our Facebook group (you'll need to join the group but it’s a one-click process).