Search · Oct 7-14, 2026
From page three to a plan
Search Console showed impressions with almost no clicks. I treated it as a ranking problem, rewrote the page Google was already testing, built the collection page for that query family and cleaned up what search engines were told to crawl.
about 420impressions in 28 days
1click from search in the same period
24average position, which is page three
Diagnosis
Search Console showed the store was already being found: one family of queries (women's sherpa clogs) produced most of the impressions, and one product page carried more than half of them. With almost no clicks at that position, the problem was ranking, not indexing. That decided the work: strengthen the pages Google was already testing instead of publishing more pages.
GA4 told the same story from the other side: only a handful of organic sessions out of roughly nine hundred, with the rest in a Direct bucket that behaved like automated traffic.
What changed on the top page
| Field | Before | After |
|---|---|---|
| SEO title | Sherpa Fleece Fur: Women's Lined Animal Print Clogs | Women's Sherpa Clogs | Fleece-Lined Animal Print Clogs |
| Meta description | Keyword-stuffed and cut off mid-sentence in the result | States the prints, fur colors, sizes, made-to-order and free US shipping in one sentence |
| Opening copy | A marketing line about elevating a shoe collection | A heading with the searched phrase and a factual paragraph: material, lining, heel strap, vents, prints, shipping time |
| Image alt text | A 310-character keyword-stuffed string on three images | One line: product, lining, print |
| Internal links | None to a category page | Plain-text link to the new collection page from two products, plus the homepage hero button |
Protected on purpose: the product title and slug (changing them would also change feed titles and links), prices, variations and the existing FAQ markup.
A new collection page
I built and published a collection page for the query family: a heading, a short intro, a product grid, a "which clog" list, and sizing, shipping and care sections with a link to the men's pair. A duplicate H1 in the draft was removed before publishing.
Catalog-wide method
- Export the whole catalog, including every SEO field, as the restore point.
- Generate candidate focus phrases from each product's own name, preferring phrases of three or more words that are unique across the catalog.
- Score each candidate with the SEO plugin's real analyzer and keep a rewrite only when it raised the score and the phrase already appeared in the copy.
- Write the fields through the store's REST API, leaving status and URL untouched.
Result: 214 of 348 products rewritten, average score about 24 to 66. The 134 left alone had no usable phrase, no unique keyword, or no score gain, and were reported rather than forced.
Clean index, honest markup
- Cart, checkout, account and duplicate payment pages noindexed; the page sitemap went from 46 entries to 20 real pages.
- When the sitemap froze at a snapshot, I proved it by comparing 45 published posts against the file and documented the cut-off. It recovered the same day.
- FAQ markup only where the answers are visible. No review or rating markup until real reviews exist.
- Digital downloads: FAQ markup per product and schema without shipping or physical-return terms.
Tools
Google Search Console, GA4, Rank Math SEO, WooCommerce REST API, WPCode, and a CSV backup before every write.
Part of my 2026 revisions to abundifind.com, a WooCommerce store I own and run. All revisions · Portfolio