SEO for a Shopify store: how template descriptions swallowed 60% of the catalogue

About the project

A homeware store on Shopify. 2,400 products in the catalogue, 940 of them findable in search. The client believed the problem was links, and had been buying them for six months.

What the check showed

We started simply: took the sitemap and compared it with what was actually indexed.

A gap of 1,460 pages. Then we looked at how those pages differed from the indexed ones. They differed in one thing: the product description.

The store used a description template per category. “Wooden chair, natural materials, comfortable shape, delivery across Ukraine”, and then the same for thirty chairs, with only the name and colour changing.

To a search engine, thirty pages with identical text are not thirty products. They are one page and twenty-nine duplicates it chooses not to show.

What we did

Rewrote descriptions for the 600 best-selling products. Not all 2,400: first the ones producing 80% of orders. Each description answers three questions the template does not: what situation this product is for, how it differs from its neighbour in the catalogue, and what is wrong with it for some buyers.

That last one matters. A description that says who the product is not for gets read for longer and returned less often.

Removed filter duplicates. The catalogue generated a URL for every filter combination: colour, material, price. That produced over 40,000 pages with near-identical content, and they were eating the crawl budget. We closed everything to indexing except two combinations people actually search for.

Linked products to each other. Every product page got a “goes well with” block linking to three related products. That is internal linking and an upsell at once.

What worked

The biggest jump came not from rewriting descriptions but from closing the filters. The index started growing nine days after that change, before any new text existed.

The crawler stopped spending visits on 40,000 empty pages and reached product pages that had been sitting in the sitemap for eighteen months.

What went wrong

We closed the filters in one go and accidentally closed two pages that were bringing traffic: “white chairs” and “extendable tables”. We found out three weeks later, when traffic on those queries went to zero.

We brought them back as standalone pages with their own text. They recovered within a month, but the month was lost.

What we should have done

Before a mass closure, export the list of pages with at least one search impression in the last 90 days and exclude them from the list. Ten minutes of work that would have saved a month.

What a proper plan does to the economics

Before: 8,300 organic visits and 96 organic orders a month. After: 24,800 visits and 285 orders.

At an average order of $47 that is the difference between $4,512 and $13,395 a month from a channel whose traffic is not paid for. The cost of the work paid for itself in the first seven weeks.

A separate word on the links the client had been buying for six months: they were not a mistake, but they could not have helped. Links strengthen a page the search engine already sees. There is no point strengthening a page it considers a duplicate.