We Asked AI Assistants for the Best Sneakers. Here Is Why Nike Kept Showing Up

The question we hear most from ecommerce leaders is some version of: how do we become one of the brands that pulls up when people ask what the best X is? Footwear is a useful place to look, because one brand pulls up almost every time.

In March 2026 we ran a benchmark across AI shopping surfaces for the sneaker category, and Nike appeared in every unbranded question we asked. Here is what we asked, what we saw, and what the product data behind it looked like, including where Nike has gaps.

What we asked, and where

We tracked 75 shopping questions across ChatGPT, Google AI and Perplexity in March 2026. None of them named a brand. They were written the way a shopper types, grouped by persona: performance athletes, sneaker enthusiasts, deal hunters, lifestyle shoppers and general shoppers. Examples from the set:

• "I'm looking for comfortable running sneakers with visible air cushioning, what do you recommend?"
• "What are the best basketball sneakers available for over two hundred dollars?"
• "What are the best running sneakers with visible air cushioning for under one hundred fifty?"
• "I want to buy comfortable leather basketball-style sneakers, can you show me some options?"
• "Can you help me find classic leather sneakers for a street style look?"

These are category questions with buying intent. They force the assistant to choose products on what it can read about them rather than on brand recall. Alongside Nike we tracked 50 footwear brands and retailers, including Adidas, Under Armour, Reebok, PUMA, New Balance, Converse, lululemon and Gymshark. Every number below is a Paz observation from that March 2026 run.

What we saw

Nike appeared in all 75 questions, on all three surfaces. Its average position when listed was 2.8, so when an assistant recommends sneakers, Nike is usually the second or third product named. Our visibility score for Nike, which weighs how prominently a brand appears and whether it is a recommendation or a passing mention, was 7 out of 10 and trending up. The weakest dimension was source citations, where the assistant links to a page rather than naming the brand.

The gap to second place was the surprising part. Adidas, Under Armour and Reebok each appeared in well under half of the 75 questions. Below them the drop-off got steeper. In head-to-head comparisons on the same question and surface, Nike won the majority against Adidas, and when Adidas did appear its average position was second. The persona view told the same story: Nike's coverage held across athletes, enthusiasts, deal hunters and lifestyle shoppers.

The useful part of a head-to-head view is not the score. It is seeing the exact question, on the exact surface, where a competitor is named and you are not. That is a gap with a specific product information fix behind it.

Inside the product data: Air Jordan 1 Mid

To understand why Nike kept showing up, we ran our AI readiness report on one of its best sellers, the Air Jordan 1 Mid, in March 2026. The report reads what an assistant can extract from the product page: brand (Jordan), category (Footwear), key attributes such as Air-Sole cushioning and unisex styling, colourways, GTINs, materials (real and synthetic leather, canvas, textile), a 4.5-star rating across 2,186 reviews at the time, price and stock status.

The overall score was 78 out of 100, which we label Strong. The story is in the four layers.

Product mapping: 91

Does the assistant know what this product is? Nike's page uses ProductGroup and Product schema correctly, so the relationship between the base model and its size and colour variants is explicit. The brand is identified as Jordan in the structured data, distinct from Nike. Every variant carries unique identifiers for cross-retailer matching. The one finding: the category is generic. The data does not say "basketball shoes" or "lifestyle sneakers", so for a prompt about the best basketball shoes the assistant has to infer the category from other signals.

Structured attributes: 75

Can the assistant compare this product against alternatives? Colour and size variants are structured and material composition is stated, so a prompt for leather sneakers can be filtered accurately. Weight is not listed anywhere, and neither are heel-to-toe drop or other technical specs that performance shoppers ask about. When an assistant compares lightweight shoes, this product cannot be included because the fact is missing.

Attribute context: 59

Can the assistant reason about this product for a specific shopper? This is where even Nike falls below the bar. The page mentions a flat-foot benefit without saying what that means for running, basketball or daily wear. It promises traction on a variety of surfaces without saying whether that means indoor hardwood or outdoor concrete. Those are the details an assistant needs to match "best sneakers for everyday wear" against "best basketball shoes for indoor courts".

This layer also explains the 7 out of 10 visibility score. When context is thin the assistant can still include the product, because the mapping and attributes are good, but it may rank it lower or hedge. Fixing context is the difference between "Nike is a good option" and "the Jordan 1 Mid is the right choice for your need".

Product context: 80

Does the assistant have enough confidence to recommend it? The review volume and rating, clear in-stock availability and well-defined shipping and return policies all count here. One finding worth noting: a recurring pattern in negative reviews about sole durability went unaddressed on the page. Assistants read review sentiment, and an unanswered quality concern can erode confidence in a product over time.

What this means for a lean team

The category leader, with Nike's resources, scored 78 on a page most teams would call good, and its weakest layer was the one that connects a specification to a shopping scenario. Use-case context is the most under-invested layer we see, and it is the one a lean team can fix fastest, because the facts usually already exist in product specifications, support tickets and reviews.

Three takeaways from this run:

• Brand recognition is an input, not the result. What gets a product included is the information attached to it. The branded versus non-branded split shows this most clearly.
• The competitive view is specific enough to act on. The question, the surface and the competitor named instead point to a particular missing attribute or missing context on a particular product.
• Record before and after. Run the same persona questions in your category, log who appears, fix the product information behind the gaps, then run them again. That is the proof an ecommerce leader can take into a budget conversation.

Run the same questions in your category and you will know whether you are one of the brands that pulls up, and why. This is the AI visibility work Paz does for lean ecommerce teams. A Paz expert finds the gaps in your product information, you approve the fixes, the platform publishes them and tracks the result. If you want this check run on your products, talk to us.