I tried using ChatGPT to buy a pair of running shoes and the results were amazing
There's still life in retail websites, but they need to adapt fast.
I still run about 35km a week from 5k Park Runs to half marathons. All of which means I spend far too much time researching trainers and visiting specialist stores to make sure I get a good fit.
Most of my online reading takes place on Reddit, or review sites that I trust. I’ll also drop the odd message into the WhatsApp group of my local run club.
This time round I thought I’d give ChatGPT a go and compare it with the retail website of my favorite brand, Asics. The results were astonishing.
At ChatGPT I entered a moderately complex question: “Best Asics men’s shoe for trail running on wet and muddy surfaces”.
Here’s what it came back with.
Let’s break this down. First, and most importantly, the reply indicates that it understood my question completely.
Then it went on to list the features that met the needs of a muddy trail run. (I could spend an entire article discussing the importance of ‘lugs’ but will spare you here).
Next, a list of Asics trail running shoes with descriptors that include suitability for different conditions and budget. Followed by thumbnails of the shoes plus prices
Finally, a side bar calling out the best match with links to retailers, including Asics.
It’s what I call a one-page funnel, from attraction to action via interest and desire. No scrolling through promoted links, no need to skip between multiple tabs and above all trust in the results.
So what happened on Asics.com? Not so good I’m afraid. In fact it was an out and out disaster. The on-page search didn’t understand the question. Even worse, it then returned a list of shoes that had nothing to do with my trail running adventure.
Other brands were just as useless. Nike ‘couldn’t find the page’ while Adidas appeared to understand the question, but returned nine pages of 400 different shoes, a digital needle in a haystack.
In other words, ChatGPT has just turned retail on its head. Why spend hours flicking between tabs and multiple retailers when you can do it all from one page without even scrolling?
What happens next? Just ask Sam Altman. OpenAI has already launched Instant Checkout where shoppers can research and pay for products within ChatGPT. In this scenario, retailers must integrate their product data, effectively outsourcing on-page search to a third party.
Does this make retail websites redundant? I don’t think so. It will take a bit of time, but most brands will enhance and adapt their existing presence to deliver a better, more personalized experience to visitors.
For instance, hosting platforms now offer generative AI search plug ins to their customers. You still need to export product data to a third party, but at least you can offer a chatbot style search experience and better convert incoming traffic. WP Engine is a good starting place with its Smart AI Search feature.
But eventually this could mean the end of navigation bars, content areas, footers, and CTA buttons as retailers build experiences around the site visitor rather than have them navigate multiple sections and drop-down menus.
It’s rather like being greeted by a personal shopper as you enter the store rather than having to wander to the escalator and check out the floor map. No wonder malls and even some websites are already experimenting with crude avatars that have more to do with animé than acquisition. But that will change soon.
In the end, what this experiment shows is that the way we shop online is changing faster than most brands realise. ChatGPT isn’t just another comparison site or review aggregator. It’s collapsing the research journey into a single, trusted conversation.
For now, that means I’ll keep using it alongside real-world run stores, where I can lace up a pair of Asics and feel how they move. But the balance of power has shifted. If brands don’t rethink their websites and search tools soon, they risk being bypassed altogether by runners like me who just want a straight answer, not a maze of menus.







