How these retailers are building AI strategies and success in marketing
- 21 November, 2019 10:27
Shifting to evolving and intent-based customer journeys, automated product recommendations and re-engaging dormant subscribers are just some of the ways leading retailers are tapping artificial intelligence (AI) to lift their customer strategies.
Speaking on a panel at this week’s Dreamforce, leaders from Adidas, Pacers Sport and Entertainment and The RealReal shared how they’re harnessing AI in their marketing and customer engagement efforts, and how others can take those crucial first steps to doing the same.
At Adidas, AI has helped the retailer better personalise shopper experiences by devising and automating the hundreds of thousands of journey combinations customers take on the path to purchase and advocacy.
“Just think about a five-step journey: If you have 10 ways of connecting any step, you have 100,000 combinations. No human being can decide that,” its omnichannel IT director, Eduard Spitz, said. “AI enables us to create personalised experiences by thinking of that journey as something which evolves into each step. It’s powered by AI rather than by a managed journey approach. AI takes that customer’s intent in the moment and automates the journey.”
Over at luxury retailer, The RealReal, AI is taking years of human marketing expertise and testing product recommendations around which of its unique designer goods to better serve customers, director of lifecycle marketing, Sara Brooks, said.
“AI is allowing us to do this in an automated way,” she said.
At Pacer Sports and Entertainment, AI is helping with establishing trust, director of customer engagement, Alana Galardo, said. “People expect you’re going to take their data and use it to make their experiences better. We’re trying to be a brand that builds trust in what you’re doing digitally, whatever your product is, and what critical steps you have to hit,” she said.
Knowing where to start with AI
While all three retailers are well on the AI path, all agreed it’s overwhelming knowing where to start.
“There’s a misconception you’ll start doing AI and the machine will go crazy because we can’t control it. It’s a fear we have to overcome,” Brooks commented. “You need to start with one small test case and just do it. Maybe it’s a small segment, where the risk is lower; start somewhere with enough scale to see the impact so it convinces you to keep investing in.”
For The RealReal, the starting point was tapping Salesforce Einstein engagement scoring against four discrete personas, and identified an opportunity to engage more with loyalists. To do this, it leveraged pre-configured algorithms out of the box to test the hypothesis that more emails would support stronger longer.
“We were willing to test this out with large enough audiences to see impact. And it did. Picking one thing you feel you can invest in is key,” Brooks said.
Brooks’ second piece of advice is painting a clear vision for the organisation. “Just saying AI means lot of things to different people coming in with automation strategy,” she said. “You need to say ‘here’s the vision for what were able to accomplish’, such as customer journey mapping, rather than saying ‘here’s a tool we can use’.”
The third must is putting the right KPIs and processes in place. “If you haven’t built business processes and your hypotheses beforehand, you’re flying blind, as you don’t have the right KPIs or measurements. You have to have a measured approach,” Brooks said.
Pacer’s starting step was also smaller scale, looking at dormant subscribers signed up to its weekly newsletter as an opportunity for re-engagement.
“We took dormant subscribers, changed the voice within the newsletter, and used sweepstakes and retail offers to try and re-engage them,” Galardo said. “The first test case gave us a 20 per cent increase within that dormant subscriber base.
“I recommend taking something you’re doing already and using that to fuel your first use case for AI.”
In Adidas’ case, however, the first step was cultural. As a large, matrixed organisation, Spitz said it first needed look across the company and map which channels customers were engaging through, from sales to marketing and engagement, and build a cross-functional picture and appetite.
“We had to define the processes, what content, time and what area,” Spitz explained. “Our marketers were still thinking about campaigns when football season hit, and struggling to let go of control. We put it into closed group, started small and just started doing this.”
Today, Adidas is using AI to aid what Spitz called ‘concept consumer DNA’, and build a 360-view of the 300 million consumers in its database.
“We have our 360-view where core profile stored, however, there’s also lot of data stored in our data lake,” he said, citing engagement, reuse and other raw data formats as examples. “We are calculating for every consumer in our database more than 10,000 attributes with AI models… then we fetch them and put them into our 360-view so they’re usable in marketing.
“These genes are helping us understand how likely this person is to buy running shoe in next three months, or how likely was it they lied when they entered their first name. We’re then using those insights in engagement.”
Overcoming AI fear
In terms of AI’s ongoing presence in their organisations, panellists also detailed how they’re articulating its role to employees to overcome latent fear around AI automating staff out of jobs through efficiency gains, scale and resource support.
“I think about it in terms of scale,” Brooks said. “We have lot of ambitions, people want to talk to us in a one-to-one way, and we have lots of messages we want to deliver. We want to get someone to drive to our upper east side store, speak to one of our experts, or see a fine piece to purchase. Or we want to take someone who’s purchased 10 times and talk about consignment.
“AI is allowing us to do that job better. We’d have to build enormous team to support all that content, coding and scheduling. It’s about taking team we have and allowing them to work on strategy, then had production work supported by AI.”
For Galardo, starting with the business benefit and objective is critical when going into unchartered territory and AI is no different. “This sets people up for success,” she said.
“Don’t be afraid if it doesn’t work, there are lesson learned and you will get it next time. But odds are if you start with objective, you will probably have great results.”
Galardo said a key use case for AI at Pacers is helping sales teams increase customer close rates.
“We’re in the business of selling tickets primarily, it’s how we positioned usage of AI as support to getting sales reps bigger and faster close,” she said. “We created those small wins, so staff realised it’s helping bringing down the amount of calls they have to do from eight to two to get a close.”
Of course, AI doesn’t have all the answers. Spitz said he’d like to see AI help with the constant challenge of attribution of customer success across channels. For Galardo, a challenge yet to be solved is valuation of engagement and how it contributed to overall revenue.
“Scoring, optimisation tools are helping to get there. Especially aggregating those two sides into one is something we’re focused on figuring out,” she said.
Another challenge is the seasonal nature of Pacer’s business. “Our engagement goes up during season time and team performance is number one impact to revenue,” Galardo said.
“To take a moment in time and say what in this environment worked and what didn’t is a big challenge. AI works on a rolling 90 days, so it takes us 90 days into season to see some of those metrics. We’re getting good at making sure we’re keeping documentation at those moments in time.”
As for what they’re hoping for in the near future, Adidas has its sights on building a better understanding of content DNA. Spitz noted the retailer is struggling to manage and access thousands of multimedia content assets created by people across the organisation as well as in partnership with athletes and clubs.
“We are trying to log pictures and analyse at pixel level what a picture is then tag it. It’s also about insights on what’s published across the family of Adidas,” he said. “Another things for us is ‘product AI’. We are trying to find out what makes a product successful. For instance, why is one specific product successful in San Francisco? Is it because it’s the same colour as local club?”
A third potential use case for Adidas is return on marketing investment and that challenge of attribution. “We believe AI can help us find the attribution in those moments, rather than having that fixed approach,” Spitz said.
The RealReal is working to extend AI into its B2B business through assistants and campaign insights. Another ongoing challenge is items sold by the retailer as one of a kind, making personalised product recommendations in any channel an ongoing issue.
“We are looking to leverage AI to knit those similar products together, so serving up other branded bags to get you to the right one,” Brooks said.
At Pacers, the team is looking to AI to move outbound messaging to a recommended channel approach.
“We’re also investigating send time optimisation on a greater scale – we see those worlds colliding so it’s no longer going into email or social, could go any channel based on person’s behaviours,” Galardo added.
- Nadia Cameron travelled to Dreamforce as a guest of Salesforce.
Follow CMO on Twitter: @CMOAustralia, take part in the CMO conversation on LinkedIn: CMO ANZ, follow our regular updates via CMO Australia's Linkedin company page, or join us on Facebook: https://www.facebook.com/CMOAustralia.