Blog

Why AI-driven audiences are grabbing ad agencies’ attention

“Half the money I spend on advertising is wasted; the trouble is…”

Stop me if you’ve heard this one before.

When American department store owner John Wannamaker said he didn’t know which half of his ad budget was being misspent, he will have had little idea that his words would go on to tee up a vast number of articles about the importance of effective marketing. And while Wannamaker’s quote may also be one of the industry’s overused, it’s an unfortunate fact that it also remains one of its most relevant, even a full century on.

The question of how much of their ad spend is being wasted – and why – is one faced by just about any company with mass-market products. But for consumer packaged goods (CPG) brands, it’s an incredibly important one – especially now. Consumer tastes have never been more diverse, making the risk of advertising the wrong things to the wrong people greater than ever.

A big part of the challenge here is audience targeting. Unless you know for certain who you’re aiming your advertising at – and how they’re likely to respond – can be very difficult to know whether your spend is being used effectively or not.

Of the various forms of audience targeting out there today, two approaches in particular have become standard fare for advertising and media agencies with CPG clients:

  • Demographics-based
    Demographics are basic identifiers like age, gender, income, education, homeownership and other “descriptive” factors. This allows marketers to group people into targeted clusters based on common traits.
  • Behaviour-based
    Behaviours are just that; the actions that we’ve taken in the past, and the habits we’ve adopted. Behaviour-based (or “behavioural”) audiences help agencies target shoppers based on what they buy, how much they spend, and when.

 

There’s a common thread to both of these audiences, and that’s the fact they’re based on things that we already know about people. That’s not a bad thing, of course. Just because an audience is based on retrospective information doesn’t mean that it isn’t useful.

Demographics can be a great way to find shoppers who lead a certain lifestyle, for instance; customers who tend towards low-priced products, perhaps, or those who are time poor and prioritise convenience. Behaviours can be used to find people who might be relevant to a specific brand, like those who buy a competing version of a product, or shoppers who have “lapsed” and no longer buy the things they once did.

That kind of information is powerful, capable of improving the ROAS of just about any ad campaign when used effectively. At the same time, while demographic and behavioural audiences can tell us a lot about how people act now, they don’t tell us a lot about how they’re likely to act in the future.

That’s where AI-driven, predictive audiences come in.

 

  • What will your customers do next?

    If demographic- and behaviour-based audiences help us understand more about shoppers today, predictive audiences help us understand what they’re likely to do tomorrow. Obviously, that has huge implications on advertising effectiveness, which we’ll come onto in a moment. For now, let’s take a quick look at how predictive audiences actually work.

    At the heart of a predictive audience is data. Specifically, information including:

    • How often someone shops.
    • How frequently they buy certain products.
    • How engaged they are with different brands.
    • How much they spend in different categories.
    • And a range of supporting demographic information.

    Happily, all of that data usually exists already; any retailer with a good customer loyalty programme probably has that information stored. Even better, when the right kind of AI-powered data science is applied to that data, it becomes possible to score customers based on how likely they are to do something – make a purchase, say, or switch to a competing product.

    Let’s look at that in practice. Say we have 100,000 different customers, and we want to know which of them are most likely to buy our product. By following the process above, we can rank all 100,000 of those shoppers, putting those most likely to buy at the top and those least likely to buy at the bottom. That’s helpful, because it allows us to focus more of our ad spend on those shoppers at the top end.

    Compare that with a behaviour-based audience. Here, we might be able to see that 60% of   that audience shops the category that we care about. That’s great, but that’s also all we know; as a result, we probably have to spread our ad budget across evenly across all 60,000 shoppers to maximise our chances of success.

    With a predictive audience, on the other hand, we know for sure which customers are most likely to buy. Instead of splitting the budget based on broad assumptions, we can instead focus it on the 10 or 20,000 (or however many) that have the greatest chance of delivering results.

    And it isn’t just purchases that we can predict either – it could be churn, trade-up, trial, or any number of other possibilities.

  • Smarter targeting, better planning

    So why exactly are these audiences grabbing the attention of the ad agencies that we’re speaking to? I think there are three reasons:

    1. Predictive audiences bring an additional level of rigour to planning

    Whenever a CPG ad agency pitches an idea to a client, they need to be confident in that proposal. Today, that goes beyond having confidence in the creative or the selected channel; it means having certainty about being able to reach the right people, too. Predictive audiences eliminate guesswork and remove assumptions, giving agencies confidence about the results they’re forecasting.

    2. They’re a ready-made solution to the problem of cookies disappearing

    There have been so many delays to Google’s planned deprecation of Chrome-based cookies that it’s easy to forget that it’s happening at all. Right now, though, that 2024 date doesn’t seem to be shifting – which means that ad agencies will need a reliable and effective way to target audiences online. Predictive audiences offer that capability.

    3. The precision they offer delivers better returns

    You can have the best creative, the most innovative campaign idea, and the most appealing product on earth – but if all of that is aimed at the wrong audience, it can all be for nothing. Predictive audiences give agencies the ability to be incredibly precise about who they target; in doing so, it helps them to use their budget more effectively, and deliver stronger ROAS for their clients.

What will your customers do next?

If demographic- and behaviour-based audiences help us understand more about shoppers today, predictive audiences help us understand what they’re likely to do tomorrow. Obviously, that has huge implications on advertising effectiveness, which we’ll come onto in a moment. For now, let’s take a quick look at how predictive audiences actually work.

At the heart of a predictive audience is data. Specifically, information including:

  • How often someone shops.
  • How frequently they buy certain products.
  • How engaged they are with different brands.
  • How much they spend in different categories.
  • And a range of supporting demographic information.

Happily, all of that data usually exists already; any retailer with a good customer loyalty programme probably has that information stored. Even better, when the right kind of AI-powered data science is applied to that data, it becomes possible to score customers based on how likely they are to do something – make a purchase, say, or switch to a competing product.

Let’s look at that in practice. Say we have 100,000 different customers, and we want to know which of them are most likely to buy our product. By following the process above, we can rank all 100,000 of those shoppers, putting those most likely to buy at the top and those least likely to buy at the bottom. That’s helpful, because it allows us to focus more of our ad spend on those shoppers at the top end.

Compare that with a behaviour-based audience. Here, we might be able to see that 60% of   that audience shops the category that we care about. That’s great, but that’s also all we know; as a result, we probably have to spread our ad budget across evenly across all 60,000 shoppers to maximise our chances of success.

With a predictive audience, on the other hand, we know for sure which customers are most likely to buy. Instead of splitting the budget based on broad assumptions, we can instead focus it on the 10 or 20,000 (or however many) that have the greatest chance of delivering results.

And it isn’t just purchases that we can predict either – it could be churn, trade-up, trial, or any number of other possibilities.

 

Smarter targeting, better planning

So why exactly are these audiences grabbing the attention of the ad agencies that we’re speaking to? I think there are three reasons:

1. Predictive audiences bring an additional level of rigour to planning

Whenever a CPG ad agency pitches an idea to a client, they need to be confident in that proposal. Today, that goes beyond having confidence in the creative or the selected channel; it means having certainty about being able to reach the right people, too. Predictive audiences eliminate guesswork and remove assumptions, giving agencies confidence about the results they’re forecasting.

2. They’re a ready-made solution to the problem of cookies disappearing

There have been so many delays to Google’s planned deprecation of Chrome-based cookies that it’s easy to forget that it’s happening at all. Right now, though, that 2024 date doesn’t seem to be shifting – which means that ad agencies will need a reliable and effective way to target audiences online. Predictive audiences offer that capability.

3. The precision they offer delivers better returns

You can have the best creative, the most innovative campaign idea, and the most appealing product on earth – but if all of that is aimed at the wrong audience, it can all be for nothing. Predictive audiences give agencies the ability to be incredibly precise about who they target; in doing so, it helps them to use their budget more effectively, and deliver stronger ROAS for their clients.

The latest insights from our experts around the world

EBIT: the next milestone in retail media measurement?
Where will our data live tomorrow?
Winning with private brands, the Customer First way
customer first data science analytics & machine learning services
Ready to get started?

Speak to a member of our team for more information