Search traffic from buyers who had never heard of the brand

Business context

A handmade ceramics brand sells mugs, bowls and dinnerware through a Shopify store, with a team of six. Most sales came from Instagram and from customers who searched the brand name directly. The owner had tried adding keywords in a hurry a year earlier, which made pages read awkwardly and changed nothing in search. The owner wanted buyers who had never heard of the brand to find the shop through search, and asked for Shopify SEO as a four-week project with monthly reporting after that.

The problem

Collection pages were the main reason people could not find the shop. Each one had a single image and a product grid, with no words explaining what the collection was for. Titles were duplicated across similar pages, so "Ceramic Bowls" and "Stoneware Bowls" competed with each other in search.

Filtered URLs were also being indexed. A search for one color could return twenty near-identical pages, which split the ranking signal and made the site look untrustworthy to Google.

Product pages were stronger than collection pages, but nobody reached them first because the searches started at the collection level. A buyer looking for a dinner set lands on a category page, not on one bowl.

The store had about 40 collection pages, and most were thin. Only six ranked anywhere.

What we built

We started with keyword research by product type, not by the words the brand liked best. Buyers search for "handmade ceramic mug" or "stoneware dinner set", so we mapped collections to those terms and merged six thin collections that had no reason to exist apart.

Each surviving collection got a written title, a short description that explains the material and how to choose, and links to two related collections. We kept the copy plain and useful rather than filling it with phrases.

On the technical side we fixed product structured data so price and availability were read correctly, set canonical tags on filtered views, and added noindex to the combinations that only changed sort order. A clean sitemap went back to Google Search Console. We tracked rankings in Semrush and revenue sources in Google Analytics, then compared the two.

Writing collection copy

Descriptions run between 60 and 120 words. Each one answers what the material is, how it wears day to day, and what pairs with it. We avoided the common habit of repeating the title in the first line, because that adds nothing a reader or a search engine can use.

Where two collections were genuinely close, we merged them and redirected the weaker URL. Every redirect points to the collection that now carries the term, so nothing is lost.

Rollout

• Week 1: Keyword research and a full crawl of every collection and filtered URL.
• Week 2: Merging thin collections and writing new titles and descriptions.
• Week 3: Schema, canonicals and index rules, tested against a staging theme.
• Week 4: Publishing, submitting the updated sitemap, then watching coverage in Search Console daily.

Before and after

Item	Before	After
Collection pages	40, mostly thin	21, each with written copy
Titles	Duplicated across pages	One target term per page
Filtered URLs	Indexed	Kept out of the index
Product schema	Incomplete	Price and stock read correctly

Results

Ninety days after publishing, non-brand clicks were 3.4 times higher. Twelve collection pages ranked on page one for their target term, up from zero. Non-brand pages went from 19% to 41% of organic revenue. The change came slowly at first, then more sharply once Google recrawled the merged collections.

The brand still gets most of its traffic from its own name, and that is fine. The point of the work was to add a second door.

Lessons learned

• Fewer, better collection pages beat more thin ones. Deleting six pages felt risky to the owner. It did not cost a single sale.
• We spent the first week on filtered URLs and nearly missed that the duplicated titles were the bigger problem. Cleaning up indexing was easy once we stopped treating every URL as equally important.
• The brand's own words for its products were not what customers typed. The owner's phrase "everyday ware" had almost no search volume. "Stoneware dinner set" did. The technical SEO work we did for a local plumbing company had the same lesson in reverse: fix what already ranks before writing anything new.