Same Mum. Different Customer.

The same busy mum walks into the same supermarket on two different days. On a Friday, she is browsing for ideas, filling a trolley for the whole family coming to dinner and she wants to cook something special: a proper joint of meat, three bottles of wine picked off a list she looked up on her phone that morning, side dishes chosen to go with each other. She’s with her son who is distracting her. Her son, Freddie, keeps putting things in the trolley and she finds this highly annoying, but this is her favourite supermarket and she loves the fresh produce. The following Tuesday, she is in and out in four minutes with a meal deal, and she’ll grab whatever ‘oven-bunger’ comes within arm’s reach. Then she’ll throw back that sandwich at her desk, and the only thing that mattered is that it didn’t take longer than 90 seconds to pay for.

Same person. Looks and thinks the same. Likes the same things and she waited in same queue at the checkout. If you are honest about it, she’s not the same customer at all. Traditional segmentation allocates this busy mum into one pen portrait, say, ‘premium, family-focused, time-poor.’

On that Friday, the moniker is true enough.

On the Tuesday it completely wrong and your data is flat as a pancake. It doesn’t tell you why she picked the cheapest sandwich over the nicer one two shelves along, or why she would have walked out if the same queue had been much longer. A segment built around who she is has nothing useful to say about what she needs on a Tuesday lunchtime. Worse still, being too narrowly focused on who she is averages two very different customers into a boring, inactionable segment.

I have spent 26 years building segmentations just like this. Their goal is to ensure marketers aren’t targeting a nebulous blob but tailoring propositions to meet divergent needs and comms to resonate with different audiences. Segmentations are the lifeblood, the glue that binds everything together… Do pick your analogy. Yet too often they gloss over reality: that a single way of grouping customers does not work. And the more one thinks about it, the more nonsensical and counterproductive the idea becomes: that it is possible to summarise all these different shoppers by attitudes, behaviour, needs and geodemographics. True, that was possible in the 1960s.

But it’s not possible now, and not in a digital world. It fails every single time because the idea itself is a non-starter.

Working with demand spaces doesn’t solve the problem, but it takes several important steps towards acknowledging and mitigating this risk. Instead of grouping people by who they are, it groups occasions by what is happening: who is present, what context they are in, and what need is live for them at that moment. Plainly: not just who, but who, where, when or why, all four, applied to the moment rather than to the person. Under that lens, ‘dinner party for six’ and ‘solo lunch under a deadline’ are not two moods belonging to the same person. They are two entirely separate demand spaces that our exampled Mum happens to pass through both in the same week.

And this is freeing. A single product and a single message aimed at ‘her’ will end up serving the ‘average’ of two occasions that don’t actually average well together: something too expensive to buy on impulse on a Tuesday, and too plain to serve to six of her beloveds on a Friday. Building for the two demand spaces separately, one fast, cheap and guilt-free for the lunch occasion, one genuinely impressive and easy to serve for the dinner occasion, wins both moments instead of half-winning an imaginary customer who exists only in the segmentation deck.

None of this could come from asking her which space she is in. She doesn’t care; she’s not a marketer – and asked directly, on either day, why she buys what she buys, she will likely give you the same answer both times: she cares about eating well, about value, about her family. That answer would be sincere, but almost useless, because it describes how she thinks of herself, not what predicts her choice come the moment of truth. Because there are several of them and that’s why Demand Spaces are built the other way round. First, a statistical model is run across real purchase behaviour to work out which contextual variables, time pressure, mission, occasion. They predict what sorts of things are bought. Only once those drivers are proven does the clustering happen, grouping occasions around the variables shown to matter rather than the ones a respondent, or a researcher, assumed did. In this very simple example, we’ve got two spaces, ‘Weekend shop’ and ‘Midweek top-up’. Mum appears in both.

What comes out the other end is a map of those spaces, each one scored on how important the need is to the customer and how well anything on the market currently meets it. The Tuesday lunch occasion fast, cheap, still wanting to feel like a decent meal is a crowded, badly served corner in most categories: everyone competes on price. Okay, we’ve got something here…

And this is the freeing bit. Acting on it means deciding which occasions are worth building for, judged commercially on revenue potential and honest fit with what the business can do with it, and then deciding what to build, which propositions, which brand in a portfolio, against that occasion’s specific unmet needs. What typically is developed for the Tuesday occasion is rarely a new product. It could be different packaging, a faster till or a price point that matches the moment, instead of a customer who tends towards ‘premium’. The Friday occasion might get nothing more than a recipe card next to the wine. Same Mum. Same attitudes, with the same perception of the chosen supermarket.

So none of this can start or finishe with who the customer is – a Demand Space Segmentation quantifies customers and their commercial value on what is actually happening. Keep pretending everyone is defined by their mission alone, uses the product in the same way all the time, or has the same price sensitivity no matter the day of the week, and you keep targeting customers who aren’t there.

This way of thinking about the spaces the business serves, isn’t only about getting an in-store aisle just right. The same thinking applies wherever the same person with the same product does a different job with it: a phone for work email and for family photos, a car on the school run and a schlep around the M25. If the occasion changes what is needed, the person is the wrong thing to be segmenting. And sure, you may want to shortcut and segment just on needs. Just needs. That’s what is it boils down to, right? But then you forget the person: that Freddie is 40 years old, has a family of his own and the roast dinner is at his house. He’s choosing because he’s paying, and that’s why she’s annoyed, and you just lost your brand, comms, positioning and any idea of what is actually happening. Same Mum. Different customer. Wasn’t kidding.