When Nike released this shoe last year, it sold out online within minutes. How did it get so hard to buy sneakers?
Welcome to the bot wars.
The sneaker craze began nearly four decades ago, with the debut of the first Air Jordan.
Back then, sneakerheads who wanted to get their hands on the latest styles had to do so in person.
Limited-edition shoes, many of them designed in collaboration with statusy street wear brands, would command long lines outside shops.
As the value of these rare sneakers rose, high-profile releases became more chaotic at stores, and sales began to move online.
Brands aggressively rolled out themed drops, such as these New Balances, inspired by John F. Kennedy’s love of sailing.
Then Kanye West entered the game, turning shoe-buying into a social media frenzy.
Digital releases created demand for automated checkout software — i.e. bots — that would enable some shoppers to buy up supplies of rare shoes, then sell them at a premium.
These days, bots are increasingly sophisticated, but so are the defenses retailers are implementing to combat them in this high-tech arms race.
BOSTON — When Bodega, a streetwear shop in the Back Bay neighborhood of Boston, released a hyped, limited-edition New Balance 997S sneaker in 2019, the entire stock sold out online in under 10 minutes.
There was one problem, though: About 60 percent of Bodega’s sales went to shoppers gaming the system with bots, timesaving automation software used to speed through checkout. The bots had claimed hundreds of pairs of New Balances for a single customer; many other shoppers failed to secure just one.
“We got destroyed by bots,” said Jay Gordon, one of Bodega’s owners. “It was making it impossible for our average customers to even have a shot at the shoes.”
Shoppers armed with specialized sneaker bots can deplete a store’s inventory in the time it takes a person to select a size and fill in shipping and payment information. For limited-release shoes, the time advantage afforded by a bot could mean the difference between disappointment and hundreds of dollars in instant profit.
In the case of Bodega’s New Balance drop, one person managed to buy a pair of the $160 sneakers before the product page was even live. Others seemed to navigate the site with superhuman efficiency, zooming from product page to purchase confirmation in 30 seconds.
Though Bodega had limited each shopper to a maximum of three pairs, the store found that it was about to ship 200 pairs of New Balances to several addresses in the same apartment building in New Jersey.
To most customers, bots are the bane of online shopping. But for sneaker brands and retailers, the relationship is more complicated.
Thanks to resale sites like StockX and GOAT, collectible sneakers have become an asset class, where pricing corresponds loosely to how quickly an item sells out. Sophisticated sneaker bots, which can cost thousands of dollars, are key to creating the artificial scarcity that makes a sneaker valuable and, in turn, makes a brand seem cool.
It all raises a big, difficult question: If the bots lose, who wins?
Status Sneakers and the ‘Swoosh Curve’
For most of the 20th century, sneakers were considered little more than utilitarian footwear for sports. But that changed dramatically in the 1980s, in large part because of Nike’s Air Jordan. The original Jordans came out in red, black and white — so defiantly different from other sneakers at the time that the National Basketball Association fined Michael Jordan for breaking its “uniformity of uniform rule.”
Michael Jordan wearing the original Nike Air Jordans during the 1985 N.B.A. All-Star Slam Dunk Competition.
Andrew D. Bernstein/NBAE via Getty Images
The original Jordans and subsequent models ushered in a new era. Sneakers were no longer bland shoes with extra padding and rubber soles; they were fashion accessories and expressions of identity. It wasn’t long before sneakers became a collector’s item starting in the early 2000s when Nike released Dunks, originally a basketball shoe, in limited quantities at independent skateboarding shops, such as FTC in San Francisco and Uprise in Chicago.
Nike often collaborated with skaters, designers and streetwear brands such as Supreme, which elevated the SB (for skateboarding) Dunks into a status symbol. Each release had a unique look, back story and catchy nickname that made the shoe feel more exclusive. For example, the so-called Tiffany dunks featured a turquoise color that resembled the boxes of the famed jeweler.
Over the last decade, most major sneaker brands have turned to high-profile collaborations. Kanye West worked with Nike and Adidas on realizing his vision for Yeezys. Nike teamed with Virgil Abloh’s Off-White to put a new spin on popular shoes from the company’s archives. Nike also tapped the design sense of Travis Scott for more than a dozen pairs of shoes since 2017.
These days, there are highly anticipated drops almost every weekend. It is not unusual to see a handful of big releases — usually coming from Nike’s SNKRS app — in a week. In online discussion forums, every new release is dissected like a company going through an initial public offering.
“It’s more about how much would you make off those shoes versus what would you do in those shoes,” said Nick Engvall, a footwear consultant and founder of Sneaker History. “We’re becoming flooded with the idea that sneakers are a commodity and that we should always be thinking about the dollar amount.”
StockX, a popular sneaker marketplace, found that most hyped releases follow a similar pattern: the “Swoosh Curve,” named for Nike’s trademark logo. When a new shoe is announced, its resale estimate is high but decreases steadily as the release date nears. The price hits a low when sales begin, then climbs steadily until it plateaus and completes the curve.
The ‘Swoosh Curve’
Jesse Einhorn, a senior economist at StockX, said the Swoosh curve reflects supply-and-demand dynamics and ultimately the upward pressure on sneaker prices as fewer unworn so-called “deadstock” or sold out pairs remain.
“While prices do fluctuate significantly around the time of release, the long-term appreciation tends to be steady and consistent,” Mr. Einhorn said.
When the pandemic hit, sneaker resale reached a frenzy on sites like StockX and GOAT. Rare shoes benefited from a lockdown-fueled investment mania that pushed up the prices of cryptocurrencies, sports trading cards and even real estate. The sale price for a new pair of vintage “Chicago OG” Air Jordan 1s from 1985 went from $3,000 in 2017 to $7,500 in May 2020 to $19,000 in February, according to StockX.
Bodega Meets Its Match
It wasn’t like this when Bodega opened in Boston in 2006. The store had no website, so anticipation for major releases was built in person, said Mr. Gordon, who owns the store with Oliver Mak and Dan Natola. Sneakerheads would travel from New York and Montreal and wait in long lines to get the latest design.
Bodega’s Los Angeles shop.
About 10 years ago, the owners thought it was becoming unsafe to have shoppers camp out overnight in front of the store, so major releases moved online. It was “before the bots arms races,” said Mr. Gordon, so Bodega allowed people to buy sneakers online on a first-come, first-served basis.
By around 2015, the site had 20,000 people appearing for major releases even though they only had a few hundred pairs of shoes. It didn’t take long before the bots swooped in. Bodega started offering web raffles, but people deployed bots for that, too. Employees had to manually check each winner so no one was securing an unfair share of shoes.
For years, Bodega put up with the bots and did what it could to mitigate their impact. But after the 2019 New Balance release, Bodega decided it needed to be more proactive or it risked losing ordinary customers who felt that the game was rigged.
Bots are not illegal, nor are they exclusive to the sneaker industry. They are used to obtain anything in high demand with limited supply. During the pandemic, people amassed stockpiles of video game consoles, graphics chips and even children’s furniture using bots. For Shopify, the Canadian e-commerce giant behind dozens of the buzziest sneaker boutiques (including Bodega), protecting against a bot onslaught is a part of keeping sites up and running.
Shopify’s Defensive Team vs. a Bot-Making Teen
The face of Shopify’s bot defenses has been Jean-Michel Lemieux, a plain-spoken Canadian engineer who was, until recently, the company’s chief technology officer. His public antagonization of bot users — who are also known as botters — has made him something of a hero among sneakerheads.
“I actually don’t like sneakers. I like computers,” Mr. Lemieux said in an interview earlier this year. “I’m a sneaker celebrity without being a sneakerhead, mostly because I’ve got to protect their platform. I know more about bots than maybe anyone on this planet because I had to reverse engineer them to understand how they work.”
Jean-Michel Lemieux, Shopify’s former chief technology officer, with his shoe collection at his home in Ottawa, Canada.
Dave Chan for The New York Times
Shopify uses different techniques to prevent bots, including puzzles and trivia questions that are difficult for an automated bot to solve. It has also taken steps to prevent transactions when a shopper’s checkout path follows the shortcuts used by bots.
Shopify’s job, he said, is to slow down bots to a human level. In the last few years, Shopify has devised custom, one-off defenses for retailers who want to stamp out bots from spoiling their major releases. In March, Mr. Lemieux gleefully tweeted a video of botters lamenting the difficulties of cracking Shopify’s custom bot protections.
Dennis Ho, a senior product manager at Shopify focused on bot protections, said that his team working with retailers tries to change tactics every time.
“We understand that nothing we do will be bulletproof forever,” said Mr. Ho, whose team will carry on Mr. Lemieux’s work.
If there is a person who keeps Shopify employees awake at night, it’s probably Lucas Titus, a 19-year-old who started college in London this month.
Mr. Titus is the founder of Cybersole, one of the best-known sneaker bots. It works on a wide range of retailers including Shopify stores, Supreme and sneaker chains like Foot Locker and Finish Line.
Similar to the sneakers it helps to procure, Cybersole sells at a significant premium. The company has limited licenses to its bot to 5,000 users. While the official retail price for a license is around 300 British pounds, or $420, it sells for 10 times as much at resale.
Lucas Titus, the 19-year-old developer of Cybersole, at his home in London.
Amara Eno for The New York Times
Mr. Titus said he started reselling sneakers when he was 14. Early on, he found success with using computer software to simulate multiple smartphones to game a raffle run by Adidas to secure four pairs of Yeezy sneakers. Mr. Titus resold the shoes, pocketing a profit of 1,000 pounds per pair, he said.
“I realized that automating things was the best way to secure not just one pair but multiple pairs,” Mr. Titus said.
He experimented with other technologies and taught himself how to code. He wrote a basic automation script to submit 50,000 entries into a sneaker raffle. Soon, sneaker buyers started encouraging Mr. Titus to sell his work. In 2018, he started Cybersole, which gained notoriety as one of the few bots to work on Shopify.
That year, the bot was put to the test when Nike released an Air Max 1/97 in collaboration with Sean Wotherspoon, a famous sneaker collector. Nike had allocated shoes for Kith, a sneaker boutique in New York, Los Angeles and Tokyo, to sell on its website, which is powered by Shopify.
Cybersole users, exploiting a shortcut in Shopify’s checkout process, cleared out Kith’s entire stock of 700 pairs, Mr. Titus said.
“We got every single one of them,” he said. “That really piqued interest from tons of people in our products and we also got the attention of Shopify.”
The Pros and Cons of a World Without Bots
After the Nike Air Max 1/97 release, Mr. Titus said Shopify began to implement anti-bot measures at checkout. It only took a couple weeks before he and other bot developers found a workaround.
Mr. Titus said the bot has successfully completed two million automated checkouts, or transactions worth around $300 million since it went live in 2018. That’s to say nothing of the millions more it’s allowed resellers to rake in as profit.
Cybersole’s interface as Mr. Titus uses the software to purchase sweatshirts from Kith.
A bot alone is no guarantee of success. Many prominent botters run multiple types of bots for major releases, because each one has different strengths and weaknesses. Some botters rent dozens of computer servers in the same facilities as the retailers to save milliseconds on data latency.
Mr. Titus said he understands the frustration some sneaker buyers may feel about bots. But as he sees it, he is merely providing tools that people want. That includes retailers who have seen bots generate demand for their products.
“While they have to act like they’re trying to stop bots, it’s making them a huge profit,” he said.
He has developed a friendly rivalry with Mr. Lemieux. When Mr. Titus showed off a new Tesla, Mr. Lemieux suggested that he decorate the car with some Shopify decals.
Mr. Lemieux said Mr. Titus “is a very, very good programmer” and that he would try to hire him for Shopify if he were willing to cross over to the other side.
“I’ve had some nice interactions with him,” Mr. Titus said. “Obviously we can’t disclose too much of what either of us are doing because it’s kind of a cat-and-mouse game between Shopify and us.”
Ahead of a special release, the New Balance 990v3 to celebrate Bodega’s 15th anniversary, the boutique and Shopify had devised a few obstacles to slow the bots down. The first was to place the product on a brand-new website with an unguessable address — analogwebsitewrittenonpaper.com.
Right away, botters saw this as a deterrent.
“We may have to go manual,” said Trevor Roskovensky, a sneaker buyer, in a YouTube video of him trying to buy the shoe live.
Bodega also added a question that shoppers had to type out the answer: What are the last four letters of the alphabet? After that, shoppers had to complete a challenge of drawing boxes around airplanes.
Shopify’s bot obstacles include typing out the answer to a question and drawing boxes around airplanes.
This was intended to throw a wrench into the store’s usual checkout procedure and make it difficult for anyone to automate the process. And it seemed to be working when the shoes went on sale in June; Bodega and Shopify didn’t see much bot activity. But then, they started seeing very little activity at all.
Shoppers started to encounter error messages as they tried to pay for the shoes.
For the next 30 minutes, only a trickle of orders were being processed. Many potential buyers gave up, assuming that the shoes were probably sold out already.
Shopify said later that routine system maintenance, unrelated to the flood of orders coming in, crimped its capacity to process payments.
For Bodega, the damage was done. Because of the payment issues, it took 45 minutes for the shoe to sell out completely, which for a major release is an eternity.
“If a shoe like this doesn’t sell out in less than 15 minutes, it’s considered a failure,” Mr. Gordon said. “It’s really disappointing because we did all this to give the ordinary customers a shot.”
The slow sellout time didn’t seem to go unnoticed by the resale market. Even though most of Bodega’s previous New Balance releases carry a significant premium to their retail price, the 15th anniversary shoes are selling at close to retail on StockX.
It’s possible that if Bodega took no steps to curb bot activity, the store could have sold its entire stock of shoes to botters before the problems kicked in because of how quickly bots complete transactions.
“It’s depressing to think about,” Mr. Gordon said. “At some point, you have to ask, ‘How much time are we supposed to spend to stop people from buying our products?’”