The majority of retailers agree AI is the future of the sector, but many businesses are struggling to implement it. Retail Week examines why
According to a recently published survey by Capgemini, 74% of retailers are currently investing in complex AI pilots but only 1% of those reach full-scale implementation.
This figure shows retailers are struggling to get to grips with AI, despite all seemingly agreeing that it is the future of the industry.
The question then is a simple one: why? The answer, as with so many of the issues facing the sector, is far more complex.
“They’re doing some cheap pilots, a lot of the time, for no other reason than to get themselves marketing headlines”
Patrick O’Brien, GlobalData
Bhavesh Unadkat, principal consultant at Capgemini, says many retailers want to invest in AI but few know why.
“A lot of retailers come to us saying they want to explore and invest in AI solutions, but when we ask them what they want those solutions for, they don’t really know. There’s no point picking AI for the sake of it,” he says.
“Too many people just think that AI is a kind of magic dust that’s going to solve everything, but that just isn’t the case.”
Even making this first step is complicated because it is hard to differentiate between what is true AI and what isn’t.
True or false
For Patrick O’Brien, UK research director at GlobalData, true AI refers solely to machine learning and goes beyond simple algorithms. He believes many UK retailers are simply announcing the fact they are exploring AI solutions to generate positive PR.
He adds: “They’re doing some cheap pilots, a lot of the time, for no other reason than to get themselves marketing headlines and give people the impression they’re doing something exciting in AI when in reality they’re doing nothing of the sort.”
However, Julian Burnett, IBM vice president of distribution in the UK and Ireland, argues AI is a very “broad church”, ranging from everything like chatbots and onsite search and navigation technologies, to systems that can automate currently human-conducted tasks, such as merchandising, pricing and even HR functions.
“Retailers go to pilot, but they’re piloting the wrong things… so they decide to kill the pilot before they get to scale”
Bhavesh Unadkat, Capgemini
He says he would be very surprised if most retailers weren’t already implementing some iteration of the former.
“The websites of retailers have been exploiting algorithmic decision making for years,” Burnett says. “There are onsite search and navigation technologies from a range of different organisations that employ AI to particular levels to give you recommendations, different search results depending on different criteria, tracking your journey. It’s already very prevalent.”
Without diving too deep into the semantics, most agree that solutions a layperson would consider ’artificial intelligence’ still have the potential to revolutionise the way retailers operate, and could save millions – if not billions – of pounds in the process.
In the UK, Marks & Spencer recently launched a new image recognition platform called Style Finder, which allows customers to upload photographs of clothes to its website. Users are then directed to similar items on its ecommerce platform.
While it seems that only the biggest retailers with the deepest pockets can afford the time and money to innovate, there are plenty of solutions available for those with more limited budgets.
Cost versus ROI
The cost of implementing complex, large, full-scale AI solutions is expensive. Combined with the sector’s collective ‘fail fast’ attitude towards innovation, this means, says Unadkat, that many retailers are killing off AI projects before they have come to fruition.
He says while most retailers begin an AI project with the intention of easing “customer pain points”, moving away from the age-old measurement of return on investment is difficult, particularly given the uncertainty of the current market.
“Retailers go to pilot, but they’re piloting the wrong things,” Unadkat says. “They don’t see that customer engagement, so they decide to kill the pilot before they get to scale.
“Other times, they’ll pilot something in a way that is purely ROI focused which, conversely, fails to address any customer issues and so fails anyway.”
“It’s not 100% assured that what you spend your money and time on will achieve the outcomes you’re seeking”
Julian Burnett, IBM
Another dilemma for many retailers is how long they should keep pumping money and time into a project that hasn’t achieved success right away. Killing off a perceived failure quickly is tempting, but Unadkat believes many retailers aren’t giving AI solutions a chance to work over the long term.
Burnett agrees, saying implementing large-scale technological change is difficult for any business, let alone at retailers, which by their very nature have an “operational imperative” to drive day-to-day sales.
“After all, it’s not 100% assured that what you spend your money and time on will achieve the outcomes you’re seeking,” he says. “Many retailers I’ve spoken to about AI have had to make very conscious choices about making investments in areas where they need certainty, versus those where they can experiment.”
Burnett points to examples of grocers using “visual and colour recognition” trained AI, which can “look at the deterioration rates of groceries as they improve through supply chains” and can be used to determine everything from sell-by dates through to promotion periods.
He also says a number of retail HR departments are using an AI solution called Hire View, which “augments the recruitment process using visual, verbal and audio-based analysis to do the first screening of interviews via video”.
Pair up with partners
O’Brien believes the only answer for many retailers looking to move forward with AI is to partner with specialists. He believes UK retailers should approach large technology companies offering partnerships with a “gainsharing element” so that both can equally “share the risks and the rewards”.
“If these big technology companies really want to engage with retail, and retailers really want to see results, that’s the only way it can work,” O’Brien says. “Otherwise it’s just going to be the very biggest players that win out.”
For retailers looking for a less expensive and more simple solution, Unadkat says there are plenty of relatively cheap, off-the-shelf technologies available for those who want to “dabble” in the AI space. However, those looking for something more sophisticated are going to have to invest heavily.
It all comes back to the same old conundrum for retailers: investment versus return. But as the sophistication and availability of AI technologies continue to increase, retailers are going to have to adapt and change with them.