An AI Helped Me Find Running Shoes for the NYC Marathon. Here's How It Worked
With my first marathon less than 100 days away, I gave artificial intelligence a go in hopes of finding the perfect trainers.
Finding the right running shoes can be tough. Can AI help?
Like a lot of other runners, I obsess over shoes. Compared with other sports, running doesn't require a lot in terms of equipment, but you can't cut corners when it comes to your feet.
For me, a good fit and comfort are most important, but I also don't want shoes that will slow me down. Super-cushioned sneakers might be great if you're doing a loop around the neighborhood with your friends, or if your job requires you to spend all day on your feet, but not when you're trying to cut a few minutes off a race time.
That search for the perfect combination has felt like a never-ending quest since I started running a couple years ago. Now, training for my very first marathon, the TCS New York City Marathon on Nov. 5, the stakes are higher than ever. So when I was offered the chance to try out Fleet Feet's new and improved shoe-fitting software that's powered by artificial intelligence, I went for it.
But that doesn't mean I wasn't skeptical about its capabilities. Up until recently, a lot of consumer-facing AI has been more hype than reality. Meanwhile, I've been shopping at Fleet Feet, a national chain of specialty running stores, since shortly after joining my neighborhood running group in March 2022.
For more than a year, the company's in-house shoe nerds, which Fleet Feet refers to as outfitters, have largely kept my feet happy. They've answered all of my nitpicky questions and their recommendations changed as my running needs and goals evolved over time.
How does AI play into that?
In this case, AI provides a way to let store employees quickly compare the specific dimensions of my feet with those of millions of others, along with the designs of the shoes in their inventory, to pick out which ones might fit me the best.
The AI isn't designed to replace expert employees, it just gives them a better starting point for finding shoes with the correct fit, says Michael McShane, the retail experience manager for the New York store I visited.
"It turns the data into something much more understandable for the consumer," McShane says. "I'm still here to give you an expert assessment, teach you what the data says and explain why it's better to come here than going to a kind of generic store."
Anyone who's ever set foot, so to speak, in a running store knows there are lots and lots of shoes out there, and everyone's feet are different. What could feel like a great shoe to one person, could be absolute torture to run in for another.
A look at some of the data collected by a Fleet Feet Fit ID scan.
Originally rolled out in 2018, Fleet Feet's Fit Engine software analyzes the shapes of both of a runner's feet (collected through a 3D scan process called Fit ID) taking precise measurements in four different areas. It looks at not just how long a person's feet are, but also how high their arches are, how wide their feet are across the toes and how much room they need at their heel.
Plates in the scanner also measure how a person stands and carries their weight. Importantly, the scanner looks at both feet. Runners especially put their feet through a lot of use and abuse, making it likely that their feet will be shaped differently,
Mine were no exception, One of my feet measured more than a half size bigger than the other. I can't say I was surprised. In addition to ramping my training up to an average of 20 miles a week over the past year, my feet have also suffered through 17 years on the mean streets of New York, two pregnancies and one foot injury that left me with a wonky right big toe.
What was a little surprising was both feet measured bigger than my usual size 9 or 9.5. I've always had big feet, especially for a woman that stands just over 5 feet tall, but I'll admit that it was still a little traumatizing to be trying on shoes a full size larger than that for the first time.
The software's AI capabilities allow the system to then quickly compare the data from a customer's scan to all of the shoes in the store's inventory, as well as the millions of other foot scans in the system. Each shoe is graded as to how its measurements matched up with the customer's. Color-coded graphics show how each shoe measures up in specific areas.
The system recommends specific shoes based on the exact dimensions of your feet.
While store employees have used versions of the software including the AI over the years, Fleet Feet says the latest improvements make it consumer facing for the first time, instead of something that takes place completely behind the scenes. The ultimate goal is to add it to the company's website to make it easier to find shoes that fit online, something that's notoriously tricky even for the biggest running shoe enthusiasts.
In addition to telling McShane and me how well a shoe could potentially fit, the software gave me a specific starting size to try on, since sizing can vary depending on shoe brand and model.
And I sure did try on shoes. The AI gave McShane loads of suggestions to start with, but it was up to him to narrow it down for me, taking into account my training needs and preferences. Ultimately, I wanted something cushioned and comfortable enough to get me through a marathon, but still light and agile enough that I wouldn't feel clunky or weighed down.
I also wanted something new. After a year of almost religiously wearing Hoka Cliftons for everyday runs, they now felt too bulky and slow. I also liked the Brooks Ghost trainers, but more for walking around New York than racing.
And I was more than happy to say goodbye to a pair of Nike Zoom Fly 5 shoes that I bought for the NYC Half Marathon. Their carbon-fiber plates and light construction made them super speedy, but their lack of heel cushioning gave me monster blisters that would explode and bleed. Sure I could have taken them back, but I liked their speed so much I just tapped my feet up every time I wore them to protect against the rubbing.
The MIzuno Wave Rider 26.
I spent well over an hour at Fleet Feet trying all kinds of shoes. Since the AI had pinpointed the appropriate size for each model, the sizes I tried on varied but they all pretty much fit. That in itself was a time saver. The main challenge was figuring out what felt the most comfortable when I took a jog around the store.
A pair of Brooks Glycerin felt cushy, but also a bit clunky. I loved a pair of Diadoras from Italy, but they ran small and the store didn't have my size, which probably would have been a monster 10.5, in stock. Conversely, a New Balance model I tried seemed too roomy to give me enough support.
For me, it was about finding the right level of cushioning and weight. Per McShane's advice, I tried my best to ignore colors. When it comes to running shoes, I'm a big fan of bright, fun colors, but looks don't help with comfort or cut seconds off your mile pace.
After many, many boxes, it came down to the Asics Gel-Cumulus and Mizuno Wave Rider (both $140). Both were light and springy and I took more than one jog around the store in both of them. I also tried them out with a new pair of insoles ($55), which also were fitted to me with the help of the AI.
I've never used insoles before, but I was told that they would give me greater support for the kind of double-digit mile training I had ahead of me, improving my endurance and reducing the chance of injury. Socks are also key to preventing dreaded blisters, so I grabbed a pair of my go-to Feetures Elite Ultra Lights ($18).
After much debate, I ended up walking out of the store with the Mizunos. While I've had Asics in the past, I've never tried Mizunos before. They seemed a bit faster and more tailored to my feet than the Asics were. It also turned out that they were on sale and I ended up getting them for $105.
That's because there's a new version rolling out that the store didn't have in stock yet, so they weren't in the system for the AI to find. While it was nice to save $35, had I known that I might have gone with the Asics just because they're more current.
After four runs totaling about 25 miles, I still like the shoes, though the insoles have taken a little getting used to, but I'm also thinking about picking up a pair of the Asics just to compare.
For most people, this use of AI will probably go unnoticed, at least until it's added to the website. While officially now geared to the consumer, it still seems more of a tool for store employees. Sure, data-crunching AI can be great, but it's the efforts and expert advice of the outfitters themselves that are going to ensure that I keep coming back to their stores.
After all, the TCS NYC Marathon isn't until Nov. 5 and I've got a long road of many miles and many, many pairs of shoes ahead of me before I reach the starting line.
Fitness Equipment
Fitness Accessories
Fitness Tech
Fitness Nutrition