The Netflix approach to retail marketing

Midway through our conversation Peter Ellen pauses to read me a quote. It’s from Todd Yellin, Vice President of Product Innovation at Netflix, who said: “Geography, age and gender, we put that in the garbage heap. Instead, customers are grouped almost exclusively by common taste.”

This is part of a new way of thinking in the world of customer marketing, an approach in which Edinburgh and Paisley-based Big Data for Humans is carving a niche for itself.

‘Martech’ – marketing technology – is busy turning decades of customer segmentation learning (classifying people into groups by labels such as ABC1/C2DE which indicate the top and bottom socio-economic tiers) on its head. Instead, marketers are starting to pay very close attention to data, and how it can unlock much greater insight into their customer base.

Who would have known, for example, that for one client of Big Data for Humans – a leading department store – that their most valuable customers for fashion retailing were men? Women account for more in terms of sales volumes but the most value came from discerning male customers.

“The really interesting stuff when we do work with the clients – there’s usually some quite big surprises as to who their customers are,” says Ellen. “The problem with the idea of a typical customer is that it’s usually a stereotype. The real customer is never as trite and stereotypical as the initial assumption is.

“What we did in Big Data for Humans was that firstly the methods for understanding who your customers are were broken for lots of enterprises. So essentially, in order for an enterprise that trades across channels to build a traditional customer insights stack, they need to go and write out million-dollar cheques with big IT firms and then hire IT experts to do the job. And then often typically those projects take very long periods of time to come to fruition. Sometimes they never come to fruition.”

Ellen is the former founding MD of Fopp, the record store chain, and says his interest in customer data began when he was able to get his electronic point of sale (EPOS) systems networked, allowing the business to analyse its stock, sales and supply chain in “immense detail”. He also might be partly responsible for getting Cuban band leader Perez ‘Prez’ Parado to Number One with Mambo No 5 in the late 90s, but that’s a story for another day.

He went on to co-found Maxymiser, a firm that specialises in optimizing web and mobile customer experiences, sold last year to Oracle for hundreds of millions of dollars. Ellen is not allowed to put the actual figure on record, only to say that it was a “big exit”.

With Big Data for Humans he has just completed the first round of investment funding, raising £1.5m, with three institutions and four angel investors behind the company, including the Scottish Investment Bank (the investment arm of Scottish Enterprise).

The firm is halfway through the second round of funding and also took part in the Techstars start-up accelerator in London. With around 150 meetings in 12 weeks, the experience was full-on, but the aim is to get a year-and-a-half down the line in that time, to get an ‘unfair advantage’ over potential competitors.

The product itself, the Customer Graph, is a software-as-a-solution (SaaS) platform, which enables retailers and travel businesses to log on and build a customer marketing programme within seconds, by an automated process. The principle goes back to the idea that Ellen is keen to reiterate, that the old way of segmenting audiences into group stereotypes is no longer fit for purpose; the future is in building and understanding networks of customers who are similar to each other, who instead can be thought of as ‘archetypes’.

“What you end up with is a simple map which shows you exactly who the customers are by group and how they are connected to each other,” says Ellen. “It’s something anyone can understand and the benefit of that method is it doesn’t squash anyone into groups in which they don’t belong.”

Ellen explains that archetypes are “extreme versions” of everyone else in a network, who if understood properly can be used to determine marketing strategy: i.e. how to sell them more of something. “Let’s imagine in a supermarket I’m a weekly shopper and I do my shopping online and have it delivered to my house,” says Ellen.

“In that weekly shop I buy a range of products. In those products I might be particularly over-represented in chilled foods or might be 10 times more likely to buy that than other people. It doesn’t mean by volume I am the biggest contributor by sales but it means I’m more likely to buy them than everyone else. So then you start to learn what makes me different from other people. So if an archetype changes their behaviour that behaviour will start playing out with other customers soon. With a stereotype you won’t know that until it’s too late.”