Can an app know a customer better than a personal shopper?

Source: Lily
Jun 29, 2017
Matthew Stern

Knowing the shopper is said to be the key to sales conversion and the team behind shopping app Lily claims to have an algorithm that lets them understand fashion shoppers better. Lily purports to be able to determine the emotional connection a shopper has with her clothing through what its creators call the “Perception and Empathy Engine.”

Lily, which won this year’s SXSW Accelerator Pitch Event, initially built its recommendation engine based on over 10,000 interviews with women, as described in a recent Forbes article. The app asks a user questions via text about her perception of her body, then points her to clothing that it determines she will find flattering. Lily co-founder Purva Gupta told Forbes that the algorithm accomplishes this by matching “emotions, preferences and perceptions” about the customer’s body with clothing from her favorite stores. Nordstrom, ModCloth and H&M are among the many popular retailers that Lily users can shop through the app.

Recommendation engines have become central to the way many online companies do business. Netflix uses the technology to push users to new content. For, it offers the opportunity for the site to induce an impulse buy or upsell with no sales staff required. Amazon has also been taking the principal of algorithm-based recommendations into the physical world at its bookstores where the selection is based on what books are its biggest online sellers.

In fact, some argue that being able to offer accurate recommendations may be the defining factor in the success of top companies in the coming years. A recent article in Newsweek speculates that Apple may be culled from the top five most valuable companies in the U.S. based on its inability to leverage data to better recommend products to customers the way Amazon and Google do.

While Lily uses conscious responses to attempt to build a profile of the customer’s unconscious emotional desires, some retailers have been playing with technology that could add even more data points to an emotion-based recommendation engine.

Uniqlo, for instance, implemented an EEG reader that purports to read customers’ brainwaves to determine their mood and match it to an appropriate t-shirt that the customer might purchase.

DISCUSSION QUESTIONS: Do you believe recommendation engines can effectively read shoppers’ “emotions, preferences and perceptions” to drive sales? How important will the technology be to the success of e-commerce sites and physical store environments?

Please practice The RetailWire Golden Rule when submitting your comments.
"The more machines know about customers, the smarter those machines get at connecting data to recommendations and ultimately purchases."
"Even if recommendation engines can read shoppers as effectively as a personal shopper, it doesn’t mean that’s what the customer wants."
"Come on retailers! Stop horsing around with this kind of unproven pseudo-science, and do the right thing!"

Join the Discussion!

18 Comments on "Can an app know a customer better than a personal shopper?"

Sort by:   newest | oldest | most voted
Jon Polin

Can a recommendation engine know that an individual customer is feeling anxious and therefore she’d like a red t-shirt? Probably not. But if that same recommendation engine knows that 20,000 (or twenty million) women have predominantly bought red t-shirts after claiming to feel anxious, now that engine is smart and can make data-driven recommendations that make sense. The more machines know about customers — demographically, psychographically, emotionally, etc. — as those customers are buying select items, the smarter those machines get at connecting data to recommendations and ultimately purchases.

Brandon Rael

The algorithmic technological innovations are indeed very impressive and if you combine these tools with an actual human experience you have a very compelling combination. The recommendation engines work extremely well with Amazon and Netflix’s platforms, as things become predictive via the powerful insights consumers are providing.

The key differentiator for brick-and-mortar operations these days is that the hybrid approach to retail works most effectively where sales associates still are providing consumers with the personalization they are seeking, powered by recommendation engines. This blending of the physical and digital works extremely well with the emerging showroom retail industry (Bonobos, Warby Parker etc.)

Paula Rosenblum

RSR partner Steve Rowen found a great site yesterday. It is an AI inspirational poster generator called Inspirobot. It shows really clearly AI’s ability to detect nuance. One of the favorite call-outs is a goldfish swimming around with the screen proclaiming “Before Inspiration, Comes the Slaughter.” Very inspiring.

So the short answer is, no … a recommendation engine is not equivalent to a personal shopper.

Art Suriano

I think technology today is getting sophisticated enough that it can read customers’ emotions, preferences and perceptions to a point but not 100 percent. As for the Lily app for apparel, I believe there will be some benefits. But it’s important to remember that technology is a tool and that tool only works when the customer understands how to use it and finds a benefit when they do. Clothing is very personal, so I can see the app making recommendations, but it will still come down to the customer trying on the clothes to see how they feel about them, how the clothes fit and how they look wearing them. No app can tell a customer that, at least not yet! However, when the customer has no salesperson available to assist them an app like this could be the next best thing.

Ed Dunn
2 years 22 days ago

Henry Ford once said any customer can have a car painted any color they want so long as it is black. Recommendation engines will likely have people wearing the same clothes in the same social circles. A good recommendation should be good enough to find “hidden treasures” which are items no one else in the region purchases as well as disclose popular clothing and unpopular clothing with the recommendations.

Meaghan Brophy

Even if recommendation engines can read shoppers as effectively as a personal shopper, it doesn’t mean that’s what the customer wants. The human connection and extra effort that drives a great stylist relationship are appealing to many shoppers. I think the key is finding a balance between people and technology so that the process is efficient, but the end result is still delivered by a person.

Shep Hyken

An app that is fueled with AI can help a customer in some cases better than a sales person at a store. This only works when the app has the customer’s buying history, which has to be more than just one or two purchases. Once the app learns the likes and dislikes of a customer, it can make suggestions based on thousands of other customers’ buying patterns, habits, fashion interests, etc. That’s what AI does. It takes past data and knows how to apply it to the current situation. Now, if you combine the app with a personal shopper, you have a winning combination that combines data-based recommendations with a human-to-human connection.

Ralph Jacobson

There is no question that machine learning technologies are making effective product suggestions for shoppers already today on several merchants’ sites. The capabilities exist to capture the aspects of the shoppers mentioned in the article and more. Actually, the more data that is consumed by these artificially intelligent technologies, the smarter and therefore more accurate they get in their suggestions.

Max Goldberg

This might work well with Millennials, who seem loathe to converse with anyone, but not for the rest of the population. I doubt that a bot will soon be able to recognize human emotions.

Lee Kent

Using these engines to help on the journey is great. Making them the entire experience? Not so much! And that’s my 2 cents.

Peter Fader

Count me (among others) as highly skeptical. Three concerns:

1. The validity of this approach to (supposedly) capture emotions, etc.
2. Even if it were valid, its linkage to actual purchasing.
3. The failure of retailers to “walk before they run.” Don’t even think about doing this kind of “far out” analysis before you’ve taken full advantage of the richness of your boring old transaction data. Some of these emerging insights could possibly add some color to such analyses, but it can’t replace them. And shouldn’t precede them.

Come on retailers! Stop horsing around with this kind of unproven pseudo-science, and do the right thing!

Celeste C. Giampetro

@Peter I concur with your skepticism. For one thing, people change their minds and emotions — especially about clothing. Most don’t wear a uniform, but like variety and watch trends (or the weather) to reflect their evolving style. Having that personal connection with a stylist can help as a last touchpoint for conversion, but can also aid in discovery that the algorithm might not pick up on.

Ken Morris
Understanding your customers and personalizing their experience based on customer context is imperative to building loyalty and maximizing sales. While I am not convinced retailers can effectively use AI apps to read customers “emotions” today, there are many other elements of customer context that offer insights to personalize the shopping experience. BRP defines customer context as “the interrelated factors of customer insights and environmental conditions that make the shopping experience relevant.” With advances in technology (networks, WiFi, mobile, NFC, Beacons, etc.) retailers have the ability to access more customer information than ever before and in real-time. Retailers have the ability to know what a customer has in her closet, what she previously purchased, what she browsed on the Web site and abandoned in her online cart, when she is near your store and even exactly what she is browsing and where within the store. In addition to customer insights, customer context considers environmental conditions such as current and forecasted weather, time of day, time of year, media (news), social media, traffic, holidays, events, and other… Read more »
Naomi K. Shapiro

While AI can work with data to learn customer preferences, it’s pushing the envelope to think that AI can read shoppers’ emotions and perceptions. A “perception and empathy engine”? Really? Sounds like it was created by the same people who hawked snake oil and other cures throughout the ages.

Ricardo Belmar
AI, machine learning, and algorithms have advanced tremendously in recent years and we’re now seeing significant examples of what is possible for recommendation engines, and every aspect of a personal shopper. That doesn’t mean most customers WANT to have an app or machine tell them what to buy vs a personal shopper who understands their style. However, why couldn’t luxury brands use this technology to enhance their personal shopper skills and deliver a truly unique buying experience? That certainly would give a shopper a reason to visit their store vs a fast fashion brand. I see this as one of those technologies that certain brands will use in place of human interaction (say fast fashion, discount apparel for example) while other, high-end, image conscious brands will use this as a supplement to enhance their experience. The technology is here to stay and will only get better so why not use it to improve the overall experience? Of course the best experience will be the one where the technology is seamlessly integrated and transparent to the… Read more »
Cate Trotter

Give them enough good data from a wider enough range of shopper and recommendation engines can do a pretty good job of suggesting products that an individual shopper might like. AI and machine learning means these algorithms will keep getting smarter and recommendations will improve in turn. For someone who is trying to navigate the endless array of products available online it could be a real boon. I think though that at this stage a recommendation engine can’t replace a good real-life personal shopper in the physical store. A good personal shopper can find out all about a customer and their style, while knowing what products are in available and pull them together to find the perfect outfits. They can also offers crucial feedback on how things look (something that customers often want from someone) and even help with suggestions on how to make things even more perfect (eg with some light tailoring).

Pavlo Khliust

I strongly believe that technologies applied in measurements are way more powerful than human perceptions. You all know polygraph tests, right? Polygraph technology is a good illustration of reading and interpreting emotions. Here is the same story with a recommendation engine but much more sophisticated and complex. To surpass a personal shopping assistant (human being) one should apply not only categorizing customers based on machine learning, collaborative filtering, Big Data and complicated math models but also systems psychology. Systems psychology (the combination of theoretical and applied psychology) is a key here, and it can be digitalized, thus measured and applied.

Franklin Chu

Apps and algorithms can make better recommendations to customers and increase the chances of a sale. Retailers with big ambitions in the China market should also realize that, with more players on board, the ability to make relevant, effective product recommendations can potentially help them stand out from their competitors. What retailers can do is to provide as much consumer feedback as possible to enable such a recommendation system to work. Working with platforms that already have data on customers and similar products can also save a lot of trouble. More importantly, retailers will need an experienced local partner that understands how these technologies make the customer experience more pleasant and convenient.

"The more machines know about customers, the smarter those machines get at connecting data to recommendations and ultimately purchases."
"Even if recommendation engines can read shoppers as effectively as a personal shopper, it doesn’t mean that’s what the customer wants."
"Come on retailers! Stop horsing around with this kind of unproven pseudo-science, and do the right thing!"

Take Our Instant Poll

How likely is it that recommendation engines trying to capture customer emotions will become more popular in the next five years?

View Results

Loading ... Loading ...