Fashion
ThredUp Solves ‘Paradox Of Choice’ In Fashion Resale Using AI
Imagine you’re a retail CEO going into the next earnings call right after introducing an AI-assisted shopping innovation that you know will be a true game-changer for the business.
While it is just being rolled out to customers, it promises to unlock the value in the company business model that has laid dormant since the company was founded 15 years ago.
You want to shout from the rooftops how it will activate new customers, increase the lifetime value of existing customers, grow the average basket size, move more inventory, put distance between the competition and otherwise grow sales and increase profits.
But first, you have to deliver disappointing results from the past quarter and own up to strategic mistakes you’ve made. That’s what ThredUp CEO James Reinhart just faced.
“Looking at the numbers piece, it was pretty bad. On the other hand, we’ve never launched a product this compelling since I started the company. It’s a weird place to be, so positive and optimistic about the future while digesting the fact that it was not a great quarter,” he shared with me after the earnings call.
Bad News First
After growing revenues 12% last year to $322 million and significantly improving its adjusted EBITDA from a loss of $43.4 million previous year to a $17.4 million loss in 2023, then making even greater strides in the first quarter by reducing EBITDA loss to $0.7 million on a 5% revenue gain to $79.6 million, ThredUp took a step back in second quarter.
Revenues contracted 4% year-over-year to $79.8 million – off its guidance of $81 million to $83 million. Active buyers and orders were down, 3% and 6% respectively, and EBITDA loss was $1.5 million, an improvement over same quarter last year at $5 million, but off that achieved in first quarter.
Rather than make excuses in the earnings call, Reinhart admitted the company got over its skis by making two strategic mistakes that are being corrected.
Exiting Europe
First, its European expansion, which held so much promise upon the acquisition of Bulgari-based Remix for $28.5 million in 2021, has proven too big a nut to crack.
“Despite bringing in new leadership, upon strategic review, we determined the Remix business needs a longer-term turnaround. As such, we’ve made the difficult decision to divest our European business and have commenced seeking strategic alternatives,” he said.
Revert To Proven Marketing Strategy
Second, it changed its marketing strategy halfway through first quarter with the aim to increase lifetime value from customers through a new offer structure and retention incentives while reducing customer acquisition costs. That proved to be a mistake.
“After reducing spend and exploring some bold changes over nearly 90 days of testing and observing retention metrics, we found ourselves worse off,” he said. The result was the company acquired 90,000 fewer new customers who will not be repeat customers over the rest of the year.
The company has quickly reverted to its old marketing and retention model and Reinhart reports it’s seen “immediate recovery” in June and July.
Nonetheless, the damage is done and the company has reduced previous year-end guidance to between $298 million to $302 million in revenues on a consolidated basis, down from $328 million to $338 million. Taking Europe out of the equation, revenues should range between $247 million to $251 million.
“We’ve learned important lessons in Q2 and in Q3. Unfortunately, these initiatives, which we take full accountability for, will have a lingering impact for the remainder of the year,” he explained.
Under Promise, Over Deliver
Perhaps another life lesson Reinhart has learned is to “under promise, over deliver” concerning the launch of ThredUp’s new suite of AI-powered shopping features.
Yet he couldn’t hold himself back. “I want to emphasize that this is the most significant product launch we’ve ever had at ThredUp,” he said during the earnings call.
“This isn’t ThredUp plus some new AI experiences. This is a fundamental upgrade in how we’re innovating on behalf of the customer. We believe that AI disproportionately benefits our business relative to other marketplaces and retailers.” And I must to agree.
AI Solves The ‘Paradox Of Choice’
The single biggest problem customers face when shopping on ThredUp is there’s just too much stuff. The company releases 40,000 unique SKUs every day, yielding an overwhelming number of items retrieved in every customer search.
With a typical Thredup page on a desktop computer displaying six items across and 20 rows down, finding the right item is challenging, to say the least, even with search features like size, color, style, brand and price options. No matter how determined, few have the patience to scroll beyond a page or two.
The new suite of AI-powered tools built on machine-learning algorithms that process natural language and imagery cues solves for the ‘paradox of choice’ problem that can discourage the typical shopper and keep them from coming back.
Now shoppers can enter exactly what they are looking for in the search bar and get a selection of items that fit the bill, such as a summer sleeveless dress for the beach, a cocktail outfit for office party or a beige-plaid coat for fall.
It produces results so much better than before and by applying the standard search features, like style, dress or coat length and color, you can really zoom into what you want.
After finding something you like, you can use the Outfit Inspiration feature to put an entire look together based on any number of style codes presented.
For example, a search for a blue-stripped dress shirt gave the options to complete the look in a contemporary business casual or sophisticated artsy way with pants, shoes, handbag, even a hat or sunglasses.
And what I think is the ultimate game changer to unlock ThredUp’s potential is the image search feature. A shopper can upload a picture they like from a magazine, a display in a store window, Instagram post or any other visual and the system will process the image to deliver individual items that most closely resemble each piece.
I tried it several times, uploading pictures of models wearing what I consider exemplary quiet-luxury looks, and every time, the first styles presented were a perfect “dress for less” or more accurately, a “dress for much, much less” alternative.
“It’s building on our extensive backend operations that allow us to process so many individual items, tag, photography and price them. Now we can turn it toward the shopper to take advantage of the long-tail of the products we have,” Reinhart explained.
Proof In The Pudding
Not everyone is convinced that AI deployed toward the consumer will work.
“We’ve seen businesses implement AI to help consolidate and streamline operations, but have yet to see it be successful when being implemented to help the consumer,” observed Juan Pellerano-Rendon, chief marketing officer at Swap, a retail e-commerce technology company.
“Although this is a shiny, new feature, it is unlikely to change consumers’ shopping habits in the near term and therefore unlikely to change ThredUp’s fortunes,” he continued.
Reinhart and I disagree. And recognizing that this is new technology for both the company and the shopper, ThredUp is actively seeking advice from guests about what they like and don’t like about the new features and probing ways to improve the customer experience.
As reflected in ThredUp’s latest guidance, Reinhart isn’t promising much from this new shopping experience in the short term, but as it rolls out and customers learn how to use the new features, it will prove a great benefit to consumers wanting to be fashionably well dressed in a more thrifty and responsible way and a windfall for the company.