AI and machine learning algorithms are increasingly being used in martech – trained to recognise and predict customer behaviour, enabling brands to tailor messaging and experiences to individual customers.
But the hard work of marketing automation has to be done first, and in some cases, trust and buy-in is still lacking, according to James Glover, general manager of AI at Movable Ink.
“People have been talking about personalisation for years and it’s failed to deliver on its promise, and so [companies] start out with their mind closed to the idea that a machine could do a better job than a person at some of these tasks,” says Glover who was CEO at AI-powered content personalisation provider, Coherent Path until it was acquired by Movable Ink in 2022.
“By and large, their experience has been that it just doesn’t move the needle on competition with Amazon, right? So, it needs to be a big enough story that you can generate enthusiasm at the senior executive level in a way that for personalisation has just not been delivered on for really the last ten years,” he adds.
“You have to tackle that automation challenge. You have to slay the dragon first if you want to marry the princess, and the dragon in this case is the challenge of getting an email out the door, and the amount of labour that’s associated with it. So much of that has to be automated in order for it to work.”
Given that AI continues to dominate tech news, we spoke with Glover about the capabilities of Movable Ink’s own Da Vinci AI solution, which plugs in to email and mobile comms and aims to understand individual customer needs. We also talked generative AI, computer vision and human creativity.
AI as personal shopper
US retailer Land’s End is one company using Da Vinci to power its email program, gaining insights into what products and categories each customer is most likely to interact with in future.
“…the content that we put in front of the customer is based partly on their momentum, leveraging the direction that they’re heading, and partly on what’s the ideal version of that customer,” Glover told Econsultancy. “It’s kind of what a personal shopper would do.”
He explains how, from a marketing perspective, the technology drives efficiency.
“90% of the touches that Land’s End has to their customers are the emails that they send out each day, and so managing that experience [of those curated assets] through machine learning, and having the machine decide each day what each person should get… we think of that as the business problem that we’re solving for the marketer,” he said.
“So why does a marketer trust us with each Land’s End customer? Well, a machine can give a million different answers for a million different customers, whereas the marketer at most could have segmented into three or four buckets.”
Figuring out the link between fast-moving metrics and longer-term success
Glover uses a rather nostalgic metaphor to very simply highlight how the solution helps guide customers towards a goal.
“When they were trying to get ET back to Elliott’s house, they put Reese’s pieces in front of him. But they didn’t just think ‘does he like Reese’s pieces?’ – they thought ‘if he takes these Reese’s pieces, will he get closer to Elliott’s house?’ and that’s what our machine is trying to do. Not just feed you Reese’s pieces but get you back to Elliott’s house.”
In this sense, Glover says that the AI is not only designed to drive conversions, but ultimately contribute to the entire lifetime value of a customer.
“You know, nobody drives to a store because they got an email, but people who have great interactions with our email program do shop in the store more. So, part of what the solution is doing is figuring out what the relationship is between the fast-moving metrics – like open, click, and convert – and the longer-term metrics. They’re not usually at odds with each other – buying something today does make you more likely to buy something a year from now,” he said.
Again, Glover reiterates the bigger picture, and how the AI considers the customer in the context of their ‘ideal’ journey.
“Think again about that personal shopper – I know you like our sweaters, so I could probably sell you another sweater, but if I could sell you some jeans, then I might be improving my long-term relationship. The machine is also making those kinds of calculations. It values the conversion today, but it also has other variables in its equation that it’s trying to optimise for to make you the ideal version of what you could be for Land’s End.”
“We want to carefully measure the value of what AI is adding to the party”
Conversation about AI has significantly increased since the emergence of generative AI tools like ChatGPT and Dalle-E 2, but this has also led to a significant amount of hype, too. Sceptics, for instance, have highlighted the various limitations of tools like ChatGPT, such as inaccuracies and its tendency to produce factual errors.
Glover highlights multiple instances in which recent advances in AI, including generative models, has significantly improved Movable Ink’s offering, also helping partnering brands to generate more ROI from their marketing investments.
“In the original personal shopper’s world, the solution that’s deciding which Reese’s pieces to deal out to each and every person – that’s a machine learning solution that is continuously learning and getting better,” he said. “What the generative AI is doing is saying ‘I want to create new versions of the creative’ so that the other side of the house, the dealer, can have more options of what Reese’s pieces to deal to the customer.”
Movable Ink also uses AI to generate different subject lines at scale. Glover cites a hypothetical example of a Land’s End email campaign, sent to the customer with the subject line of ‘Great Outdoorswear from Land’s End.’
“But what if they didn’t open it, and you got zero value from the photoshoot [used in the email creative]? Well, what if AI could crack ten different subject lines?” he asked. “It’s about using AI to generate additional subject lines to get leverage out of the company investments that you’re making.”
Lastly, Movable Ink has also started to use computer vision, which Glover describes as “looking at an image, comparing it to the history of all the other Land’s End images, and deciding that this one is going to appeal to the customer.”
“We couldn’t do those things without the kind of advances in AI over the last year or so,” he said, “and that’s just three new ways in which it’s helped our business.”
Consequently, Glover is bullish on the core business value of AI. “If we do what the marketer used to do – from a business-as-usual perspective – we want to carefully measure the value of what AI is adding to the party. So, we know that it can do better than the marketer can.”
Personalisation “should be about discovery,” not reinforcing behaviour
One way that Da Vinci differentiates itself from other AI personalisation engines is that it is not ‘reinforcement-orientated’, as is often the case with basket abandonment triggers. Again, Glover likens Da Vinci to a personal shopper, and crucially, not the unhelpful kind.
“If you walked in the door and all [the personal shopper] said was ‘hey I know what you did last night, and I’d like to use that to try to sell you something’ – it would be a different experience than ‘hey, I know you’re into sweaters and we’ve got a great new sweater collection that I think you’re going to like.”
Glover suggests that the former experience has contributed to wider distrust in personalisation, with consumers often feeling like personalised targeting is a step too far or an invasion of privacy.
“If you’re marketing your story in as relevant a way as possible, that feels very different than an abandoned cart that you’re reinforcing. In the US, for example, a lot of the news that you’re getting is reinforcing what you already believe as opposed to what it’s supposed to be, which is helping you discover what’s out there from an information perspective. So, we would like to say that AI should be about discovery, it should be asking questions, and it should be showing you new stuff.”
AI needs to be “bundled with the right people, process, and technology”
Glover indicates that a lot of the scepticism surrounding AI is related to the the debate about technology versus human creativity, whereas, he suggests, the real value stems from a combination of the two.
“ChatGPT could get my teenage son a B+ or a C- on a paper, but if he could package ChatGPT with a little bit of energy, he might be able to get an A. These tools are not at the place where you can just say ‘write my essay’ and it will solve your problem for you.” However, when it comes to human effort combined with the technology, “there’s tremendous value in it from that perspective,” said Glover.
“Essentially, if bundled with the right people, process, and technology… we have already demonstrated that AI can bring business value, but I just think it requires that kind of context. And so I use the term ‘leverage’; it gives you leverage so that you can do more than you could do yourself.”
So, what about the belief that generative AI could erode human creativity? Glover suggests that the argument is fundamentally flawed.
“Human creativity might only be 20% or 30% of the equation, but it might be more than 20% or 30% of the value,” he said. “And this is my kind of general view of automation, is that everybody wants the best experience, so if that better experience requires a little bit of human labour, you’ll choose it anyway.”
“It’s like bottled tap water… the tap water is effectively free, but the marketing associated with the bottle is what’s going to make you able to charge a dollar for it instead of a tenth of a cent. The machine value will get commoditised, human value won’t, and so the humans will have a role to play in the process I think for a long time. That’s the history of what we’ve seen with technological innovation anyway.”