Most ecommerce teams I meet ask me a version of the same question. “We tried Shop the Look two years ago. It worked on the dresses. It often broke and was a huge manual tagging headache for my team. What has changed?”

Up until about 6 months ago, the honest answer was not much! Most older Shop the Look tools rely on pixel matching. You compare an image against your catalogue and pick the closest visual match. That works for a small slice of your range where the imagery is clean, the products are visually distinctive and the stock is stable. It often breaks elsewhere.

That is why most retailers end up with a Shop the Look implementation that covers their top SKUs and nothing else.

Shop The Look has become a staple tool to encourage catalogue discovery and conversion across many retailers, but scaling to more than 10-20% of SKUs is usually a huge resource challenge.

Why shoppable content is hard to scale

Many retailers have built in-house tools and acquired technologies that depend on pixel matching to work at scale. This approach uses tools like Google lens to identify an image and find the catalogue product that visually resembles it most closely. It does not understand what the product actually is, what category it sits in, what materials it is made of, or how the brand merchandises it. Two black blazers can be visually near-identical and commercially very different.

With flat, clear imagery, this approach can work for a few hundred hero shots. It falls apart at any meaningful catalogue size. The matches get fuzzier as the catalogue grows. Out-of-stock products keep appearing in shoppable looks. The merchandising team end up policing the output instead of building compelling experiences for their customers.

Most teams know the symptoms even when they have not known the cause. They describe Shop the Look as a feature they “had to keep an eye on.” That is the cost of running discovery on visual similarity alone.

What fashion-native AI does differently

Style Graph is the layer that sits underneath every Shop the Look experience we run. It learns the catalogue the way a brand actually understands it. Categories, materials, descriptions, attributes, lifestyle imagery, flat lays, every signal a merchandiser would use to decide what fits with what. It learns the brand profile too, picking up what the brand would never put together as much as what it always would.

When a shopper lands on an image, Style Graph is not asking “what does this picture look like?” It is asking “what is the brand saying with this look, and which products in the live catalogue best deliver it right now?”

That is the difference between matching pixels and matching intent. It is also the reason Shop the Look can finally extend past the hero shots and into the long tail.

StyleGraph AI ingests catalogue data, brand profile and content signals, along with model imagery. The LLM then establishes shop the look variations in seconds, which are automatically converted into shoppable output, ready for the PDP.  

Shoppable content in a constantly changing catalogue

The other thing pixel matching cannot solve is movement. Products go out of stock. New collections drop. Sale lines move every fortnight. A Shop the Look experience that was accurate on Monday can be embarrassing by Friday.

Style Graph treats inventory as a real-time signal. If the item in the original image is out of stock, the system substitutes the closest live match the brand would actually recommend, not the closest visual match. The shopper sees a look that always sells available product. The merchandising team does not spend their week patching gaps.

What this looks like in production

We see it play out across our customers.

Primark has 70 percent of a 16,000-product range online. Hand-tagging that catalogue was too much work, and a pixel-only tool would have stalled at the top of the range. AI-led Shop the Look is what makes that scale possible. In testing, Primark’s shop the look modules are now driving an AOV increase of 5% across over 2,000 new products per month. 

Similarly, Boden uses AI Shop The Look to create shoppable outfits, including “best fit” recommendations fully aligned with the Boden style guide, and including out of stock compatibility updated in real time.

Boden’s AI Shop The Look and outfitting module in action 

Scaling merchandisers, not replacing them

One thing worth being clear about. None of this work replaces a merchandising team. It scales them. Every individual decision Style Graph makes is one a senior merchandiser could make. The difference is volume. A merchandiser can run a hundred hero shots. They cannot run fifty thousand SKUs across multiple locales every week.

Fashion-native AI is what lets the team stop spending their time on tagging and start spending it on the merchandising calls that actually need human judgement.

That is what scaling Shop the Look properly looks like in 2026. Not bigger spreadsheets. Not more contractors. A different category of tool.

Download a full overview of AI Shop The Look here: [INSERT LINK TO SHOP THE LOOK 1-PAGER]

Sources and further reading

• How Primark Is Reinventing Product Discovery for 2026, StoryStream

• StoryStream's AI Shop The Look

• The New Rules of Fashion Discovery, StoryStream eBook

• Solving Fashion's Returns Problem with AI and Authentic Content, StoryStream