Big Data and the science that could change shopping

Big Data could up-end leasing strategies, unseat anchor tenants, and bring landlords closer to shoppers. Or could it? David Thame checks the numbers Freezy’s Christmas Adventure – an augmented reality game playing in 21 of British Land’s shopping centres – is about as cute as it gets. The app-based children’s game encourages families to follow a five-step virtual present hunt around the mall. The kids help snowman Freezy to overcome obstacles and gather gifts for Santa, while the delighted parents experience British Land’s upbeat vision to create ‘Places People Prefer’. It is the first of a four-year campaign partnership with virtual reality specialist Harmony Studios. [caption id="attachment_889409" align="aligncenter" width="847"] Freezy’s Christmas Adventure – augmented reality comes to shopping centres[/caption] But this Christmas, the science of retail will get harder edged. Big Data is coming to shopping centres as landlords roll out ambitious new customer nudge programmes and a raft of other initiatives. If it works, it could change shopping. Intu, Hammerson and British Land are all launching – or about to launch – pilot projects based on Big Data collection and real-time intervention. Using the existing WiFi infrastructure, augmented by paid-for data from phone companies and their own market research, they are now deploying state-of-the-art software to nudge, monitor and beguile their shoppers. And the word “their” is key to understanding what is going on. As shopping centre landlords increasingly place their own brand centre-stage, developing a strong two-way relationship with customers becomes vital. This is what Big Data is about. The UK’s retail landlords join global names such as Amazon and Alibaba, which are now also exploring what Big Data can do. Trevor Pereira is digital director at intu, where pilot projects begin this autumn. “Data is going to become pervasive in this business, although at present we’re just in the foothills – in the small scale trials to see what works and what doesn’t,” says Pereira. For intu this means using the WiFi log-ins as the base sample, which they then weight using more generic exit surveys and CACI data to ensure they fairly represent a cross-section of shoppers. Trials so far include working out the impact escalator closures have on customer flows and comparing shopping centre attendance on days with, and days without, events at adjoining facilities. At intu Braehead in Glasgow, for example, they compared the days Disney on Ice was appearing at the arena to non-Disney Saturdays, looking at customer flows, dwell times and which stores were visited. Similar studies have been undertaken at intu Eldon Square in Newcastle. Licenses for the software come in at a few hundred thousand pounds – not a fortune in this context – and uses pre-existing WiFi network infrastructure. “We already had the fibre ring-main WiFi in place, thanks to our 2013 digital strategy, and we have invested in new skills on the digital side. And scale matters – that we can do this over a large number of shopping centres makes it easier. This is still early days, but we are now generating masses of data that we can put to manifold uses,” says Pereira. Intu is already exploring the internet of things, looking at clever ventilation systems that predict when they will need repairing, and sensors in toilets that can literally sniff when they need cleaning. The next stage will be personalised real-time intervention. “If a customer seems to be tired, and to need a refresh and more inspiration after two hours shopping, is that the appropriate time to send them a free coffee or meal voucher? And if we did, can we prove it extends dwell time?” asks Pereira. EXPERIMENTS IN AI So far, artificial intelligence is mostly making itself felt behind the scenes at shopping centres. British Land is one of a handful of businesses experimenting with AI. Working in partnership with law firm Addleshaw Goddard and software house Kira Systems, the firm is now halfway through an 18-month pilot project to see if AI can improve the way it keeps track of retailer leases. Advanced machine learning software identifies, analyses and extracts provisions and other information from contracts and other documents. It has built-in machine learning models for common transactional needs. British Land’s Ben Dimson explains: “We have got about 2,000 retailer leases, and approaching them through AI is exciting because, until now, going through them to work out what is and isn’t possible across the portfolio has been slow, manual and expensive. AI means a way to prise these documents apart. AI is a way to re-enliven them.” App trials Hammerson’s group product innovation manager Kathryn Malloch is already piloting a programme that can answer some of those questions. A self-selection group of 350,000 Hammerson app users has been used to trial push notifications. Hammerson has sent 700,000 so far, and the result is a 12% increase in dwell time. It’s still early days but promising. Like intu, Hammerson has invested in back-end infrastructure to collect, digest and assess some vast and rapidly accumulating data sets. So far they have data on 22m venue visits from their 350,000 app users. Work is going on to tie this firmly into relevant content and real-time offers. Ben Dimson, head of retail business at British Land, says it is heading down a similar path: “We’re still a little way off responding to customer data in real time, but that’s my aspiration. I hope in two or three years we can do something targeted at a really personal level.” British Land is working with WiFi provider Purple to turn raw data into something useful. Data sets are being used to answer questions as varied as what outdoor summer temperature is too good for shoppers to want to be indoors (answer: 28 C) to whether anchor stores really do anchor shopping centres (by encouraging visits to other retailers) or whether shoppers go to anchors, then promptly go home? If WiFi analytics provide an answer to the last question, it could spell the end to generous anchor-tenant leases. “We can test whether what a retailer says they bring you actually matches reality,” says Dimson. Big Data could also refine leasing strategies. Matthew Thompson, retail strategist at Colliers International, says: “Using WiFi data analytics, landlords can begin to fully understand shoppers and so you can be surer who you ought to target as tenants.” Colin Brooks, managing director of outlet mall operators Realm, can also see leasing benefits. “Big Data means we are more able to understand turnover data and so can better define how we market ourselves to retailers. After all, retailers have lots of choices and the clearer evidence we can show of driving footfall, the better. Data is a competitive USP,” he says. Shopping is personal However, not everyone is convinced Big Data is going to define the future. Hannah McNamara, retail partner at Cushman & Wakefield, likes the idea of data sets that can point to prime pitches in real time, showing how they move during a week. She is interested in real-time interventions but adds: “I think people want an immersive real-time experience and a chance to step away from their phones for once. “And besides, it is very hard to monetise Big Data. Take simple examples like trying to convert footfall data into things like rental values – and look at the Pavegen kinetic pavement trials at Westfield and Oxford Street, where they are trying to monetise footfall by generating power.” She believes existing methods of data gathering are still valid: “The old school trick of watching people, looking at their faces, what bags they hold, where they go… I prefer that to a lot of data. Shopping is personal and you can’t reduce it to an algorithm.” Similarly, not every landlord is climbing aboard the Big Data bus this Christmas. Grosvenor insists it is quite happy with existing turnover data matched with marketing focused on weekend events. Director Miles Dunnett says: “We’re looking at real-time data at Liverpool One, but we’re not there yet.” Some may be concerned about being too intrusive but Hammerson’s Kathryn Malloch says: “Millennials want something back in return for their data. If you do that, they are quite comfortable. “The point with data is that it’s there to help our customers make their visits more valuable and enjoyable. Data has to drive customer experience – and if it does, it’s worth doing. But not otherwise.” GLOBAL TRAILBLAZERS Global giants such as Amazon and Alibaba are investing heavily in big data. The new Amazon Whole Foods will combine a bricks-and-mortar grocery retailer with advanced retail data science. This will involve optimising customer purchasing, dynamic pricing and machine learning/AI. Alibaba has set up a task force to integrate data, online, offline and logistics into a single value chain. New York-listed Alibaba announced the move amid claims that big data could be used to reshape manufacturing. SKILLING UP FOR THE ERA OF BIG DATA From data analysts to copywriters, the era of big data is going to mean landlords will have to acquire new skills. Intu has created a 50-strong digital team to cope with both incoming data and analysis, and outgoing response to customers and retailers. “We have had to move into totally new skill sets for a landlord,” says intu’s Trevor Pereira. British Land is also investing in skills. A five-strong data analysis team is now at work, drawing on a variety of specialist skills outside the usual landlord tool kit. READING CUSTOMERS’ EMOTIONS Would you be spooked by cameras hovering in the racks of this season’s frocks? Would you mind if the shopping centre CCTV monitored your emotional mood? Kate Nightingale, head consumer psychologist at Style Psychology, has worked with Harvey Nichols and Debenhams, and says sensory brand strategies are becoming popular – meaning how things feel gets more important. And one of the best ways to work out what feels right is big data about people’s faces. “We are looking at LED lights that include cameras and biometric strips to gauge shoppers’ emotional reactions. This can include emotional-recognition software using CCTV images, all of which registers the response to items on sale. “So, for instance, you touch the piece of clothing – which shows some interest in it – but then what do you think? The software can show whether you touch it with pleasure, or not – and if not, that is something the retailer wants to deal with. Maybe tell the product buyers not to use that fabric next season.” However, intu’s Trevor Pereira reckons its vast footfall would generate too much emotional data for a quick and meaningful response. “This is not a near-time focus for our research,” he says, but it’s not off the agenda. Main image: © Carsten Leuzinger / imageBROKER/REX/Shutterstock