Ecommerce Content Strategy: Fix the Category Page Before You Write Another Guide

Your category pages are working like a cashier who cannot answer a question. A shopper lands on one, finds nothing that decides the purchase, and goes back to the search results. That return trip is invisible in almost every store's reporting: it looks like a normal session that ended, not a bounce. Meanwhile the publishing schedule fills with guides that send people to a page never built to receive them.

The Blog Is Not Where the Money Is

In Similarweb's 2025 benchmark of the top 1,000 US ecommerce sites, covering January to November 2024, organic search produced 23.56% of traffic. Direct produced 57.66%, and paid search 6.88%. I am reading those numbers through SEO Sherpa rather than Similarweb's own report, so treat the shape as reliable and the decimals as directional.

Organic is a quarter of the traffic, paid a fifteenth. Publishing more content is an odd place to start.

Read that benchmark again and it gets more useful. Branded and direct visits together make up 71.7% of ecommerce traffic, and only 3.5% of the audience visited a single ecommerce site. Nearly every shopper was comparing, with another tab open. Comparison happens on a category page, a product page, or a comparison table, because that is where price, availability, and specification sit next to each other.

Search for this topic and you get numbered lists of content types, all answering what to publish and none answering which page to fix first. A blog post can start a comparison. It cannot finish one. That is the whole case for an ecommerce content strategy built around page types rather than around what you feel like writing. If your plan is full of guides and your category pages are untouched, you are building the cheapest part of the funnel. The website content strategy that decides what each page is for is the layer above this one.

The Three Page Types That Carry Ecommerce Search

Three page types do nearly all the commercial work in a store. Category pages hold a collection, product pages hold one item, and comparison pages hold the choice between items, on their own URL or as a block inside a category page.

Most stores have all three. Look at your own: two get managed, and the comparison layer is usually nobody's job.

Diagnosing a Thin Category Page

Thin gets used loosely. On a category page it means the page has products on it and no argument. Before you commission a word of copy, check four signals.

The first screen is a control panel. Filters, sort, a result count, a grid, and nothing telling a shopper what the collection is for.

The only unique text is the product names. Take the titles out and nothing remains that is not already on the product pages.

It answers what is in the category, never which one to pick. Someone who knows they need a waterproof jacket is looking at sixty of them with no guidance.

The subcollections follow the merchandising system, not a decision. A filter combination promoted to a URL is not a category. If nobody can name the shopper who wants that exact slice, the page has no question to answer.

If three of the four fire, treat the page as a rebuild.

The Baymard Institute's Homepage and Navigation research, updated in October 2026, reports that 61% of desktop sites and 65% of mobile sites score mediocre to poor on Homepage and Category Navigation, based on more than 17,000 performance scores from over 180 leading US and European ecommerce sites.

That number is easy to misuse, so here is the boundary. It measures navigation UX, not how much text a category page carries. Nobody measured content thinness in that study. What it does tell you is that the shelf is hard to walk, and a shopper who cannot navigate a collection will not read your introduction to it.

This is why a content audit comes before the rewrite, not after: you want the pages failing on structure before anyone starts writing.

The Guide That Eats Its Own Category Page

The most expensive failure looks like success while it happens. You commission a buying guide. It ranks. It earns links. Six months later the guide and the category page are trading places for the same query and neither is climbing. Nobody penalized you: two of your URLs answer the same question, so the search engine picks one per query, and the pick moves. The trade calls this cannibalization. From inside the business it feels like a page that used to work.

The fix is assignment rather than deletion. Decide which page owns which question, then make the split visible in the copy.

• The guide owns what comes before the choice: what kind of thing you need, what the terms mean, what people get wrong.
• The category page owns the choice itself: which of these, at what price, with which attributes.
• The comparison layer owns the gap between two items: this one or that one.

Then link with intent. The guide sends its commercial anchor text to the category page from inside the paragraph where the shopper is ready to move, and the category page links back only from a buying-help block low on the page.

One rule keeps the split honest. If you cannot say out loud which of the two pages a query belongs to, you have not designed a pair. You have published twice.

Content Only a Store Can Publish

Some content cannot be copied by a publisher, an agency, or a competitor blog, because it does not live in a publishing schedule. It lives in your operations.

Return reasons. Measurements per size, per product. Compatibility lists for parts, accessories, and consumables. Care instructions written by whoever handles the complaints. Fit notes from the fitting room. Which two items get bought together, and which two always come back together. Warranty terms with the excluded cases named.

Unglamorous, and the only content on your site a competitor cannot rebuild from a search result. A guide on how to choose a backpack can be rewritten in an afternoon. The note that this model's hip belt runs small, and what to order instead, cannot.

The order is rows before prose. Every attribute you capture at product level becomes a comparison column, a filter value, a fit note, a support answer, and a fact inside an article. One structured field, five uses. That is the discipline behind content repurposing tools, applied to data instead of articles.

Most of that belongs in the comparison table, the fit note, or the support flow, not in a blog post. And a wrong fit note is worse than none: it produces a return.

Where UGC Fits and Where It Does Not

Reviews are the most over-trusted asset in ecommerce content. What makes them work is where they sit.

On a product page they answer the doubt that stops the purchase. Does it run small. Did it survive a winter. That placement belongs next to the thing being doubted, not at the bottom of a long scroll.

On a category page a review block usually costs you. It adds a second scanning task to a page whose only job is comparison, and pushes the comparison information below the fold. The exception is one curated line inside the comparison table, attached to the attribute it supports.

Then there is the threshold, which most guides skip. Below a certain count, a review section argues against you: it displays how few people bought this, and a near-empty block reads as a warning. Above it, the same block becomes the strongest proof on the page. The threshold is the count at which a shopper stops reading the number and starts reading the reviews, and it differs by category.

Open three product pages and read the last review on each. Where it is still arguing about whether the product is any good, the count is not doing its job yet. Where it has moved on to a narrow doubt, the block can carry the page.

That test takes an hour and it settles an argument that otherwise runs for weeks: whether to chase reviews or rewrite the page. Below the line, rewrite the page. Above it, ask for the review, because a hundred honest reviews will beat any introduction you commission.

A Ninety-Day Sequence

This is the order I would run, and it is deliberately not the order most content plans use. Three blocks, one gate between each.

1. Days 1 to 30, commercial pages. Take the ten category pages carrying the most revenue, not the most traffic. For each, write the four things it is missing: what the collection is, which attributes decide the purchase, how the items differ, and where a shopper goes if none fit. Build the comparison layer for the top three pairs in each collection. No new guides this month. You are finding out whether your problem is content or navigation, and you cannot tell while both are broken.
2. Gate 1, end of week 4. If impressions moved and clicks did not, the snippet was the problem and the copy did its job. If nothing moved, check indexing and internal links before writing a single guide. More rewrites will not fix a page nobody crawls.
3. Days 31 to 60, the product pages that carry the returns. Sort products by return rate rather than revenue. The top twenty are where missing content costs money, because every return is a customer who did not get the information the page owed them. Add fit notes, compatibility lists, and the care detail that stops a second wrong purchase.
4. Gate 2, end of week 8. Compare return rate and support ticket counts on those twenty products against the previous period. If returns did not move, the content is not specific enough. Generic size guidance does nothing. A measurement does.
5. Days 61 to 90, the support layer, in this order. Comparison blocks on the category pages that earned them. Then buying guides for the questions your support inbox answers most often, each linked to a category page from the paragraph where it earns the click. Then distribution, because a guide nobody sees is a cost, and a content distribution strategy turns a published page into a page that gets read. Hold the review-led blocks until a product is past the threshold from the previous section.
6. Gate 3, end of week 12. By now you should be able to name, for every guide you published, the category page it feeds and the query it owns. If you cannot, the guide was a scheduling decision, not a commercial asset.

Guides are not worthless. A guide sends a shopper somewhere, and if that somewhere is a grid of tiles with no argument on it, you paid for the click and lost the sale. Fix the destination first, and your ecommerce content strategy stops being a publishing target and starts being a route through the store.