Looking for something a little outré?
Let us know.
Martine Farrugia
Written by

Martine Farrugia
Unfortunately we can’t predict the future. But using predictive modelling techniques, we can predict how customers will respond and behave to marketing.

We’re bombarded with more and more information every day, making it harder for brands to make themselves heard.

But when you start to use predictive modelling techniques, you can start to find out why people do the things they do. It’s about tapping into their deepest desires and primal instincts to predict what they’re going to do next. Which helps you create things that produce specific reactions.

We work in partnership with Bob Raleigh, founder of PathSight Predictive Science, to design creative that builds on that basis to get your customers (no matter their mindset) on your side.

It’s all about clusters

We’re all individuals. But PathSight’s research suggests that there are 18 key clusters or ‘groups’ of people, who think along similar lines.

But these groups can then be grouped into five broad clusters, each reflecting specific characteristics. Each group has certain shared likes and dislikes, and will respond to stimuli in much the same way.

These can then be grouped into five broader clusters, where each cluster represents different characteristics, and different likes and dislikes on an instinctual level. It’s a bit different to the persona model, where you rely on commonly understood ‘trends’ among different demographics. There, you might think of 18- to 35-year-olds as movie streamers, and 65-year-old males as retired golfers. But these models can be quite reductive, and rely on stereotyping to get by.

Cluster map
A graphic representation of the clusters based on inherent reactions.

That’s not to say they’re not helpful. They are, when you’re look across key demographic segments. But they never tell the whole story. Predictive modelling techniques bring all the ‘quiet’ data to the surface. The data that’s sat in datasets for years but has never surfaced. We’ve been missing out, because it’s incredibly valuable. It tells us all about the causes people believe in, their political affiliation, traits and the values they prize above others.

The insights are based on research involving over 80,000 consumers. And the resulting system means we can produce messaging which is ideal for the person receiving them.

With modular design and composition platforms readily available, there’s no reason not to tap into this. Appealing to the different clusters in different ways means you can produce better results, boosting your clickthroughs, conversions, and sales. All from using different words and images for different clusters of people.

Sounds good, right?

Making the most of it

The first question to answer is, “Who am I talking to?”

From there, you can think about where they’re looking. Then about what to say to them. When you’ve got all that in shape, your audience will get your exact point, and feel like the message was just made for them. Which it was.

This helps to build a much deeper connection between your brand and your customers.

Flipping the script horizontal diagram
Here’s how outré flips the script to improve the messaging flow.

Who uses predictive modelling techniques?

Predictive science is getting pretty popular these days.

Why? Because the brand speaks with their language. It shows them facts the way they want to see them. It makes them feel like they belong, which builds trust and keeps them coming back.

We’ve worked with PathSight to boost a number of clients’ numbers. Take the New York Jets, for example. Fans might be loyal, but they don’t think exactly the same way. So we saw an opportunity to boost the team’s email engagement. PathSight took a deep dive into fan data, and we used their output to speak directly to certain groups of fans. Open rates and clickthroughs shot up by an average of 37% for each one.

But it’s not just about words. Design’s just as important.

And for The Economist, we created three different campaigns to suit different audiences. By delivering personalized experiences beyond the expected, we made sure we spoke directly to the audience, increasing engagement and subscription numbers.

How does it work at a project level?

There are three things we need to think about.

  1. A model. We’ll examine key customer data and use this to identify the key clusters making up your audience. From this, we can start to think about their key motivations.
  2. An engagement plan. This is the strategy that helps us put everything into action, from narrative, to creative, and a bundle of words, themes and images that’ll strike the right chord with specific clusters.
  3. We’ll roll out the approach into your comms. From simple emails to portals, apps or websites – anywhere you need to boost engagement.

And the great thing about predictive science is that it looks to the future, not the past.

Traditional models are based on what came before. Predictive modelling lets you look at improving existing, and creating new, touchpoints that boost engagement in the long run.

We don’t have a bunsen burner but we can definitely set your communications alight.

Say Hello

Share this

More like this