Does Your Catalogue Breadth Work Against You?

Catalogue drift doesn't happen on purpose. A store launches with a tight proposition, usually one the domain name spells out. Then a supplier offers an adjacent line at decent margin. A repeat customer asks whether you stock something nearby. A slow quarter makes a wider net look sensible. None of those decisions is wrong on its own, and five years later a domain named for one specialist thing is carrying a catalogue that says something else entirely. The question isn't whether that's bad. It's whether anyone actually decided it.

The name makes a promise

Search engines and AI systems build a working model of what a domain is about, assembled from the products, the collections, the content and the links pointing in. Google doesn't publish a topical-authority score, and we're not going to pretend to know the weighting. What we can say from the pattern across audits: stores with a coherent catalogue and content stacked around one category get found and recommended for that category, and stores with a scattered range end up competing on every front at once, against specialists narrower than them and against generalists with a hundred times their scale. The specialist name doesn't carry the unrelated lines, either. Nobody's mental model of a store named for coffee gear includes yoga mats, and the machine version of that mental model is no more generous.

Inside the store, the costs show up more quietly. Collection pages thin out, because nobody has time to merchandise forty categories properly. Product data quality drops as range grows, because writing four hundred good titles is a different job to writing eighty. We've covered both halves of that problem before, in "Your product titles don't say what the product is" and "Shopify categories and category metafields: the unglamorous work behind AI visibility", and range growth quietly undoes exactly the work those pieces describe. Content has the same problem. A blog that once went deep on one subject either goes generic to cover the new range, or goes quiet.

AI recommendation makes the stakes higher, not lower. When an assistant is asked for the best place to buy something specific, the safe answer for it to give is the recognised specialist, and the model's idea of who that is comes from the same coherence signals. A store that reads as fifty things at once is hard to recommend for any one of them. The generalists who survive that filter are household names. Everyone else gets found by being unmistakably about something.

"Everyone else gets found by being unmistakably about something."

Breadth is a strategy when somebody chooses it

This is not an argument that every store should cull back to a tight core. One-stop breadth for a defined customer is a real position. Everything for the reef aquarium is coherent at three thousand SKUs, because the customer is the organising principle rather than the product type. The test is never SKU count. It's whether a line deepens the proposition your name makes or just sits beside it. An adjacent range the same customer buys on the same trip strengthens the store. A range that arrived because the margin looked good and the rep was persuasive mostly dilutes it.

That leaves three deliberate options, and all three are legitimate. Widen the proposition properly, which may mean the name and positioning change with it, not just the catalogue. Narrow the range back to what the domain promises, and take the short-term revenue hit with your eyes open. Or keep the breadth but structure it as clearly subordinate: navigation, homepage, collection order and content weight all pointed at the core, with the adjacent lines present but never competing for the store's identity. What doesn't work is the default path, where breadth accumulates and the site keeps presenting itself as the specialist it no longer quite is.

Pull the numbers before you decide

Export twelve months of sales by product. Tag every product one of three ways: core to the proposition in your name, adjacent to it, or unrelated. Then compare revenue share to SKU share for each bucket. The pattern we keep seeing is a long adjacent-and-unrelated tail holding around half the SKU count and a thin slice of the revenue, while consuming the photography, merchandising and data-entry hours the core range needed. The tail rarely pays for the attention it absorbs, and nobody notices, because each individual line looks harmless on its own report.

"The tail rarely pays for the attention it absorbs, and nobody notices, because each individual line looks harmless on its own report."

Search Console tells the other half of the story, and it costs nothing to check. Look at which queries you actually appear for over the last twelve months. A healthy specialist profile shows your core terms gaining impressions. A diluting one shows the profile scattering into long-tail noise, lots of queries, none of them yours in any meaningful sense.

Measure it yourself

The afternoon version, no export required: take your top 200 SKUs by revenue and do the three-bucket tagging on those alone. If more than a quarter of them are carrying a proposition your domain name doesn't make, you have a decision pending, whichever way you end up making it. Then ask an AI assistant where to buy your specialist category in Australia, and read the answer twice. Once for whether you're named. Once for whether the stores that are named carry a narrower range than you do.