A picture is worth a thousand words. And, done right, maybe a few clicks on a buy button, too.
In an interview with PYMNTS, Richard Noguera, chief information security officer at Yapstone, discussed the findings of the playbook titled “Meeting the Millennial Need For AI-Powered Visual Shopping.”
Noguera said artificial intelligence (AI) can prove to be useful at both the front end (where interaction with the consumer takes place and suggestions are presented for the next purchase) and the back end (where data is collected and analyzed).
“They are equally important, just different,” he said.
The Front End
On the front end, marketplaces seek to use AI as a tool to differentiate themselves from the competition. Amazon may stand out as one of the best-known practitioners of striving to be predictive based on consumers’ buying and browsing patterns, Noguera said.
The eCommerce giant, of course, examines frequency of purchases, cadence of purchases, and even the images of what has been browsed alongside those purchases to get a sense of what the end user is seeking — like whether the focus has been on a brand, for example, or on a specific product. Amazon’s goal, said Noguera, is to get the most relevant items in front of the user.
“Ultimately, that benefits the market space,” he said, because it might spark a consumer’s impulse to click their way through to purchase. In most retail verticals, Noguera said, the approach can be somewhat scientific, based clearly on past purchasing behavior.
In some cases, such as specialty apparel, the intent may be to view an item of clothing or a pair of shoes, and subsequently visit the store in person to try it on. In that event, the marketplace functions as a virtual catalog.
“There’s also a lot to be gained on the back end,” he told PYMNTS. Marketplaces can gain insight on how to actually present items in the future — through different types of imagery, for example — in ways that drive click rates and purchases.
Grappling With Outliers: Sweaters in Summer
Presenting the right images of the right products at the right time can be just as much art as data science.
There always are, and likely will be, outliers — the ads and product suggestions that make marketplace users scratch their heads and do a double take.
“I don’t know if there’s any one marketplace that will give you 100 percent fidelity on who you are and what you’re doing,” said Noguera. “Because when you think about the technical side of it, there is an assessment of, ‘Hey, there are others like you who bought this same item and they also bought these other things.’”
Chalk it up to the back-end data-crunching that correlates the individual’s “demographic” in terms of the time of year, the frequency of purchasing, and even whether they — and their perceived peers — have an account with that particular marketplace doing the data-crunching. The analytics sometimes have to make a stretch of assumptions to arrive at the suggestions ultimately presented to the consumer.
So, sometimes, we get sweaters suggested — in the dog days of summer.
For the marketplaces and the consumers who want that 100 percent fidelity, well, that quest starts to bump up against some privacy concerns.
As Noguera noted, for that tailored level of projection/suggestion from the marketplace, the users must open themselves up to a level of data sharing that spans what they’ve bought, what’s getting a fresh look, who their friends are, and what their friends have bought, too.
AI has its place in the subscription commerce market, said Noguera, who cited the ability to foster a sense of “mass personalization.”
“It all depends on what your relationship is with any given market space and how much you want them to know about you,” he said.
In discussing the findings of the playbook, Noguera said, age plays a role in the platform economy and in shopping visually. As might be expected, younger users, including millennials and Generation Z, rely on platforms such as Pinterest and Instagram when deciding what they want to buy.
And as younger, tech-savvy shoppers tend to have a relatively greater number of accounts than their elders, they exhibit more “digital behavior” than others and throw off relatively greater amounts of data.
“Data is the new oil of our time,” said Noguera, who added that there has been marked growth in the number of data aggregators and data analytics providers rushing to marketplace operators with the promise of revealing who buys what and what comes next (all provided for a fee, of course).
With the continuing evolution of AI and machine learning-enhanced market spaces, Noguera said the increasing reliance on visual methods of piquing buyer interest may bring product placements with social media into greater prominence. Such a shift will be subtle — and the product placements may even be subliminal.
Against this backdrop, when it comes to the collection, management and use of data, both sides of the commerce equation — the consumer and the marketplace — bear responsibility, Noguera contended.
The Consumer Responsibility
At the individual level, he said, consumers need to be cognizant of their own behaviors and examine their willingness to share those behaviors with the various sites they visit.
The overarching mindset, at least these days, Noguera said, boils down to: I only want to be known when I want to be known. “Otherwise, ‘leave me alone,’” he said.
Caveat emptor — or buyer beware — especially if some attributes stand out, anonymous or not. The comic book enthusiast who focuses on the Marvel universe should not necessarily be surprised to be presented with Spider-Man T-shirts at some point in the future. (The pool gets shallower as the interests get narrower in scope.)
The Marketplace Responsibility
On the other side of the equation, the responsibility of the market spaces is one where “once you do accept and you have that contract agreement with the end customer, you have to do everything you possibly can to ensure that people are known when they want to be known,” Noguera said.
The marketplaces also have a responsibility to vet what their partners are doing, he said (a key failing in the Facebook/Cambridge Analytica debacle).
“To me, this is not a security responsibility, but it’s just a ‘doing the right thing’ responsibility,” he told PYMNTS. “You’re being trusted with data. Don’t use that data and trust and give it away flippantly. Once it’s lost, trust is very hard to regain.”