The one thing I get most requests about these days is helping out with online audiences that are not working. And I realise I see the same problems over and over. Naturally, that’s when a blog post is born.
Most people working with marketing audiences today started long before online marketing was the norm. They still create audiences for the online world like they did (or still do) for the offline world. But the two are very different, and naturally, audiences don’t translate very well between the two contexts.
The difference between online and offline audiences is how you decide if someone is part of your audience or not.
Your audience hypothesis
Before you create an audience for your ads, you usually have an idea of who you want the reach with your product. If you’re going to market a contraception app, spending your marketing budget on women between the age of 23-45 seems fair, but if you try to sell fancy cheese, your audience is somewhat different and probably should consist of cheese lovers with enough income to spend on cheese.
Sidenote: Some people work with personas to get to know their audiences. I try to avoid that since I find it limiting. I will save my take on personas for a separate post. But its safe to say that personas create a lot of trouble when people are trying to reach their “personas” with online ads.
Let’s get back on track. Often you have to translate your business audience into an advertising audience. Maybe because you want to personalise your ads based on preferences, or age; Or, because you have a small budget and want to make sure you spend it on those who are most likely to consume it.
The limit with offline audiences
In the offline world, you have very little information about people. You often know the average income level in a zip code or a magazines rate of female readers, but you don’t have rich profiles or detailed information about a single person.
So, when you advertise offline, you do it in a zip code where the average income level is similar to what you think your audience earn. Or you choose a magazine with mostly female readers in a relevant age span. But you will never know if they like cheese or are trying to get pregnant.
What is a “proxy”?
In statistics (yes, building audiences is statistics), a proxy variable is a variable that is not in itself directly relevant, but that serves in place of an unobservable or immeasurable variable (Wikipedia).
The zip code is a proxy for income, and you can use a women’s magazine as a proxy for gender and maybe also specific interests if the magazine focuses on a particular type of content. Marketers try to find good proxies to make their advertising do better – but it is hard for some products and services.
Say you run a house cleaning service. You believe you should try to reach women because they feel more responsible for house cleaning (bleh). But you also want them to make enough money to afford your service. But how can you narrow it down further? Say you want to reach women with demanding and high paying jobs. Then you can advertise in magazines that people in this category are likely to read.
This magazine sure seems like a good proxy for your audience, but you won’t know how many of the magazine readers who are relevant to you. Some women who read the magazine might have demanding and high-paying jobs, but they have husbands that do all the housework. And others might not have a high paying job yet but wish to have it one day, so they are reading the magazine as inspiration.
Offline audience spill-over
So among those you target with an offline audience, only some people are the ones you’re trying to reach. You will have “spill over” to other groups that you’re not looking for. It’s the same with “out of home” ads, you can put them up on the bus stops in an area where the income level is high, but that doesn’t mean a significant share of those walking by, looking at your ads, won’t have that high income.
The benefit of online audiences
When you create audiences online, you have much more data about people. Either, you have your own customer data (“first-party data”), or you use data from Facebook or Google (“secondary data”), or you buy data from someone else.
Facebook and Google collect data not only on their platforms but all around the web and in apps through different scripts such as Facebook like buttons and analytics scripts. This data collection is why they know so much about their users, and this is why their advertising solutions are thriving online.
When you have data directly about each person, you don’t need proxies. You can target users on Facebook directly on income level. Or cheese interest, or their interest in contraceptive apps. These targeting possibilities are why Facebook ads can become so relevant. We don’t need to guess if those interested in cleaning services are also reading a specific type of magazine, or are more likely to have a gym membership. We can target directly towards people interested in cleaning services.
Why your Facebook Audiences don’t work
Almost everyone that contacts me, about poorly performing online audiences, have built them incorrectly. Hence, they are using proxies instead of pinpointing the behaviours or interests they are trying to target. This set-up makes their ads taking a massive detour in deciding who is relevant and who isn’t.
And the use of proxy-based audiences for social media is widespread. I’m continually meeting both media agencies and social media marketers that are doing it daily. But it is just MUCH MORE WORK that will give you a WORSE RESULT. So it’s pretty easy for me to recommend you to stop.