"I'll show you something that even the client doesn't get to see," Blair Freebairn says proudly, tapping his mousepad to spill a ream of text on to the screen. "This... is the algorithm."
In a boardroom at Pitney Bowes Business Insight's head office in Windsor, Freebairn has just unveiled the secret formula of store location. The 13 lines of code on his laptop tell retailers where to locate a branch to maximise profits - and whether or not a new supermarket or mobile-phone shop is likely to survive its first year. "Retailers today are ruthlessly business-focused," says Freebairn, who wrote the code as a variation on the Huff model, an established formula for determining the probability of a consumer entering a store. "You work out how likely someone is to want to spend their money on, say, shoes and then what the available range of shops is.
You then build a model to tell you where that customer is likely to shop, and that takes into account revealed behaviour from purchase patterns, survey data, and loyalty-card data."
PBBI programmes each store's variables into the formula. The client may want to attract particular shoppers; may have an optimal store size above which profit tends to fall away; or may need to compensate for how the brand is perceived locally. The software can also determine just how many Starbucks, say, can be squeezed into a single neighbourhood - and where they should be sited - before one branch starts cannibalising the profits of others.
Freebairn pulls up the model he made for "a major department store" to demonstrate. He asks it to calculate the viability of a new site in Brighton. "This is now going off into some custom code and running a full-blown spatial interaction model," he explains. "It's generating trip costs to Brighton from all the postal sectors within an hour-and-a-half. It then works out how many people from that drive-time match the store's shopper profile, and the relative attractiveness of a Brighton store. It then gives you the impact on sister stores."
The screen pauses as the program runs its algorithms through the data. "It tells us that a new store would do about £65m a year, which is a little low - but they probably would open that.
Bluewater would be affected by about £0.6m - that's the nearest branch." Some company models include a further step, where Freebairn deploys a "heat map" of estimated footfall to suggest precisely where a store should be.
Forty miles north of Windsor, Shelagh Herrity is using PBBI's software to locate a new Domino's Pizza store. It's just gone midday and she's in Domino's head office, a functional piece of real estate in Milton Keynes. As Herrity brings up the program on her laptop, she indicates an area around Southall, in Middlesex.
Southall isn't an obvious choice as the demographics don't seem promising. "Two of the segments we need to watch for are 'Asian Enterprise' and 'South Asian Industry'," Herrity explains. "Many only eat halal meat and they tend to cook at home. These blue dots show that there are 13,000 Asians within the delivery area." This detailed demographic information comes from financial-data group Experian, which has broken down the UK population into 61 types, and the deductions run deep. In Experian's terms, the "Asian Enterprise" type is described as having "relatively low levels of interest in style, the arts and intellectual ideas other than those that are of specific practical value". But the issue for Domino's is this detail: "A lot of food is prepared from fresh ingredients and cooked using traditional methods."
Herrity creates a map to verify whether these blunt demographic assumptions match the local reality. The computerised map correlates data drawn from two existing Domino's stores near Southall, as well as the local demographics. "This lets us predict how people in that area will behave," she says. "It shows that
'Asian Enterprise' accounted for 33,000 sales in the space of 13 weeks - so it's a good result."
She suggests that this could be because the demographic is becoming more Westernised, or because the population is largely buying vegetarian pizzas. So the area works as a potential location to support the next Domino's. Now there's just the matter of narrowing down which street.
As Domino's delivers, the franchisee needs to reach enough of its pizza-buying population within an eight-minute drive. The software computes this for an available site - and finds that the numbers work. The people of Southall shall have their store. And it will be on Allenby Road.
In the near future, Freebairn expects the science of store-siting to become far more precise. "My mobile knows which mast it's on," he says. "Suitably anonymised, that's incredibly useful information - it shows where you are and it could be linked back to your demographic. I've already seen it working in our Spanish office. "We'd be able to sit here at the computer and see people who have bought tomatoes in a supermarket, and have them flash up live on the screen. You'd need the people who own the big rich data sources - the Dunnhumbys, the credit-reference agencies, the telephone companies. They have incredibly rich data to contribute, once they can work out how they're not breaching anyone's privacy. But I think we'll have it within a year."
This is just one element within Wired UK's special report on the new hidden persuaders. You can read the introduction to the special report here and a selection of the other articles here: - Data that define your retail options
How the TV watches you
When advertising gets in your face
Mining your mobile phone logs
Your unconscious mind has already voted
Now marketing gets sniffy
Neuromarketing is a go
Eye-tracking adverts
Your secret shopping personality
Want more Wired UK magazine? Make sure you get your copy every month - subscribe online today.
This article was originally published by WIRED UK
