No thanks. I'm happy to struggle.

Blog & News

BE and segmentation - do we need both?

Category: Blog & News

“Dan Ariely points out that human beings are about 97% similar. Trying to segment your audience is important to some extent -- those 3% of differences can be profound -- but many of the drivers of behaviour are universal and not always rational, as standard economics would presume. Behavioural economics looks at the areas where that breaks down(1).”

Something I know many of my clients grapple with is how (and whether) behavioural economics intersects with segmentation.

Think of it this way. Behavioural economics is about similarities. Segmentation is about differences.

BE is about similiarities

You can survive without segmentation, but you can’t thrive without BE. Of course, the two methods of understanding behaviour complement each other and work most powerfully in tandem.

Let’s say 100 people are in a room.

A behavioural economics perspective suggests that everyone in that room is going to be influenced by the same forces as everyone else.  Forces such as:

  • Priming e.g. type of music playing and the height of the ceiling
  • Social norms e.g. who else is in the room and what they are doing
  • Defaults e.g. what drinks are being served
  • Framing e.g. what the drinks are called

By understanding these universal influences, behavioural economics gives us the most efficient way of shaping the behaviour of the group.  Our area of enquiry is not whether people will be influenced, but to what extent and in what direction.

A segmentation view suggests there are sub-groups within those 100 people, and the best way to influence them is to articulate a message that accords with specific wants and needs. We typically use proxies like age, income, attitude, gender or geography to separate and differentiate.

For our message to be effective, according to a segmentation perspective, it needs to be tailored to people who share common characteristics, with whom it will resonate. No use sending me an ad for dentures if that is not something about which I am interested.

This is where segmentation gets tricky, however, because in order to make our message efficient, we need to develop segments as large as possible to capture as many people as we can. How deep do you have to go to be effective without losing economies of scale? Writing a personalised letter to every customer might be most effective, but wildly impractical.

Introducing BEgmentation

When it comes to BE and segmentation, they are best when used together in a specific sequence. I call this “BEgmentation”

BEgmentation combines segmentation with BE

In BEgmentation, behavioural economics is the starting point, with each principle applying across the board. It’s an ‘opt-out’ model, where every principle is assumed to be in play unless there is good reason for it not to be.

Segmentation is the layer on top, where a more nuanced view of your audience can be developed. That means you could explore how segment 1 is impacted by messages framed differently, or how the same message impacts segments 1 and 2 differently, for example.

The key here is to start with BE so you know the foundations of how behaviour is influenced. If you start with segmentation, you end up having to repeat the process of working out whether a difference between groups is unique to them or something that affects all (i.e. a universal principle.)

It’s a bit like the difference between baking a large cake and icing it all at once, or making multiple cupcakes and icing each individually.

Segments of one will be a reality

BEgmentation is only going to become more important.  Traditional segmentation will increasingly be automated and personalised, with the customer’s digital paw prints scooped up and interpreted by algorithms and predictive analytics. Your role will less about defining the segments, and more about developing the pre-populated materials that can be delivered automatically.  

The only way to devise effective materials that are generic enough to scale but also be personalised is to start with a roadmap of similarities, not differences.

Imagine, you are advertising a dog food sale, for example. You devise an email to customers that, thanks to your knowledge of BE, includes a social norm. The algorithm then drops in the most effective representation of the social norm for each specific person (segment of one), increasing the odds they will respond. The email offer to me includes a picture of golden retrievers because the algorithm knows I have one, whereas my neighbour gets an email with pugs. We both click to buy.

This article also appeared in Smartcompany.

You might also find interesting: