product data

Why Fixing One Product Page Doesn't Fix the Range

Fixing one product page leaves the rest of the range untouched. See why the durable fix belongs in the catalog, not in one-off page edits.

F
Furniture Connect TeamNovember 13, 2026
Why Fixing One Product Page Doesn't Fix the Range

The first response to discovering that an AI assistant struggles with your product pages is to fix one. Pick the best-seller, add a specifications table, replace "performance fabric" with the composition, write an FAQ, state the shipping window. An afternoon's work, and the page is better for everyone who reads it.

Then there is the rest of the range. And when the fabric lineup changes, the fixed page can drift out of date.

What a page fix actually is

When you edit a product page by hand, you're typing a fact into a field on that page. The fact came from somewhere: the supplier spec, customer service, your own measurement. It now exists in two places, the source and the page, and the page copy is disconnected from the source.

That's fine for one page for a while. For a range it has two problems. Every page needs the same work, by hand, and the work is done differently each time: one writer puts seat depth in the table, another in the description, a third forgets. And every page is now a copy that has to be maintained separately. When the supplier changes the foam or you drop a fabric, each page has to be found and edited.

Across a range

Even a careful two-hour fix becomes a large manual project across a deep range: find the source, extract the facts, write them into the right places and check the variants. Teams often improve the priority products first, while the long tail remains as thin as it was.

And the priority list isn't static. Best-sellers change, new ranges arrive, and each arrival is another page to fix by hand.

Where the fix has to live

The facts an AI assistant needs are the same for every sofa: dimensions, materials, construction, durability, delivery, variants. They're the same for every bed, every dining table, every chair, within each category. That regularity is the way out.

If the seat-depth attribute is included in the sofa family, then every sofa record in that family has a slot for it. If the page template shows a specifications table built from the attributes, every sofa page shows seat depth, in the same place, with the same label, as soon as the attribute is filled. If the feed and structured data are mapped to the same attributes, they carry it too. Once those mappings exist, fill the attribute once per product and it can reach all of them. Change it once and the connected outputs can pick up the change, subject to each one publishing and refreshing; check the destination, don't assume.

That's the difference between fixing pages and fixing the catalog. Pages are copies; the catalog is the source. A page fix lasts until the next change. A catalog fix lasts as long as the record is maintained and the mappings stay in place.

What this means in practice

Three pieces of work, in this order.

Define the attributes. For each category, identify the facts that answer the shopper questions (see the questions shoppers ask AI assistants about sofas). Create those attributes and add them to the relevant family so every product has the slot.

Fill them. This is the bulk of the work and it's where tools matter. Some facts can be extracted from the supplier spec sheets you already have. Some come from a supplier's own data (a cleaning code from the fabric supplier's care instructions, not inferred from the composition). Some need a human: packed dimensions with the legs off, measured once. Doing this per record is still work, but it's data work, done once, and it can be reviewed and batched, rather than page work repeated for every product.

Point the outputs at them. The page template reads attributes into a specifications table and an FAQ. The feed and structured data carry the attributes as fields. From this point, a new product that arrives with its attributes filled can be published complete on day one, and a fact that changes is changed once, then checked at each destination.

How to tell which kind of fix you're looking at

A quick test for any proposed fix: if we do this, does the next product we add get it for free? Adding a specifications table to one page: no. Adding the table to the template and the attributes to the relevant family: yes. Rewriting one description: no. Changing how descriptions are generated from attributes: yes.

The fixes that pass the test are the ones that hold. The ones that fail are the ones that feel productive and leave the long tail thin.

The honest cost

The catalog fix is more work up front than fixing ten pages, and it touches more people: whoever owns the product data, whoever owns the site template, whoever owns the feeds. It's also the fix that can improve the website, the marketplace listings, the comparison tools and what AI assistants find, because they can all be fed from the same record. For a furniture retailer with a deep range, it's the only version of the fix that finishes.

For the full argument, including what the record needs to hold and how to check progress, see being found by AI. For a precise look at what's different about being found by AI versus traditional search, see SEO, GEO and product data.

This week

Take the fix you were about to make to a page. Ask whether the next product you add would get it automatically. If not, find the version of the fix that lives in the relevant family or the template, and make that one instead.

Ready to show and sell?

Give every team the product data, imagery and assets they need to launch faster, quote better and turn more of the catalog into revenue.

No slides. We stage one of your products live, at a time you pick.

Talk to salesOr see pricing