Category copy above the grid, below the grid, or none at all: has anyone tested all three on the same store and seen a clear winner?

1 reply

• 

  adminAdmin

  Original poster 22 hours ago

  The short answer is that I am not aware of any published, controlled test that ran all three variants (above the grid, below the grid, and none) on the same store. What is public is a handful of two-arm SEO split tests on different sites, and they disagree with each other. Removing copy from lower on the page hurt one site and helped another. Making a short intro visible on mobile helped a third.

  That disagreement is the useful part. “Placement” is standing in for three separate things: whether Google gets enough context to understand the page, whether the copy helps or obstructs the shopper, and what it costs the grid in scroll depth and page weight. On a typical store, the decision should depend more on what the copy actually says and how far it pushes products down on mobile than on where it sits in the template.

  What the published tests actually show

  All of the controlled tests I could find come from SearchPilot, which runs page-bucketed SEO split tests (groups of similar pages get the change, others act as the control). That is stronger evidence than a before-and-after chart, but each result applies to one site’s templates, audience and competition.

  Test	Change	Result	What it does not tell you
  Travel category pages (2020)	Removed below-the-fold copy that SearchPilot said “predominantly exists to target search engines, not users”	Organic sessions fell 3.8%, an estimated 9,800 sessions a month	Whether moving the copy higher, or rewriting it for shoppers, would have done better than keeping it where it was
  Retail category pages (2024)	Removed SEO text from the bottom of category pages	Statistically significant uplift from mobile, negligible on desktop, repeated in a second section of the site	Whether a short, useful intro would have beaten having nothing
  Product listing pages (2024)	Made previously hidden intro content visible on mobile	About 7.5% more mobile traffic and 7.4% across all devices, at 80% confidence	Whether the gain came from rankings or from better snippets and click-through
  Retail category pages (2026)	Cut products per page from 48 to 36	Positive result (85% credible interval), which SearchPilot read as page experience outweighing content quantity	Anything about copy directly, but it shows listing pages are sensitive to weight and length

  Two cautions on reading that table. First, 80% confidence is a lower bar than most conversion teams would accept, so the mobile intro result is directional. Second, the “removal” tests removed specific blocks of text on specific sites. The 2020 test removed copy that SearchPilot itself described as not visible without scrolling, and it still cost traffic. The 2024 test removed bottom-of-page text and gained mobile traffic, which SearchPilot attributed to added scroll depth hurting the mobile experience. Neither result tells you that “below the grid” is good or bad in general. They tell you that one site’s block was earning its place and the other site’s was not.

  Google’s guidance versus what the tests show

  Google’s clearest statements on this are older but have not been contradicted since. In 2020 John Mueller said “maybe 90%, 95% of that text is unnecessary but some amount of text is useful to have on a page so that we can understand what this page is about”, and that a list of products on top with “essentially a giant article on the bottom” can make the page’s intent harder to interpret. In 2019 his advice was blunter: “Make up your mind, focus, don’t blindly stuff text”.

  The practical complication is that practitioners see removal tests go both ways. Those observations do not contradict the guidance, because the guidance is about amount and purpose, not position. If headings, breadcrumbs, filters and product names already make the page’s topic obvious, removing a generic block loses little relevance and may improve the mobile experience. If the copy is the only thing clarifying an ambiguous category, removing it removes context Google was using.

  Two newer documentation points matter for implementation:

  • Mobile is what gets indexed. Google’s mobile-first indexing guidance says only the mobile content is used for indexing, that moving content into accordions or tabs on mobile is fine if it is equivalent to desktop, and that Google “won’t load content that requires user interactions” such as clicking to load. A “read more” that is collapsed but present in the HTML is different from one that fetches the text on click.
  • The FAQ rich result motive has gone. Many stores added FAQ blocks below the grid partly for FAQ rich results. Those stopped appearing in May 2026, so those blocks now have to justify themselves on shopper value alone.

  Why the same copy can win in one position and lose in another

  A useful way to picture it: copy above the grid is a shop assistant who greets you at the door. That helps if they say something short and orienting (“everything on this wall is waterproof, and the sizes run small”). It hurts if they give a speech while you are trying to reach the shelves. Copy below the grid is a notice board at the back of the store: most shoppers never read it, but it costs them nothing. No copy is an unstaffed shop with good signage, which works when the signs already tell you everything.

  The analogy covers the shopper side. On the search side, the mechanisms are more specific:

  • Relevance context. Google needs enough on the page to understand the category. Where it sits visually matters less for indexing than whether it is in the rendered mobile content at all.
  • Snippet and click-through effects. A clear intro can feed a better snippet, which can lift traffic without any ranking change. SearchPilot flagged this as a possible explanation for its mobile intro result.
  • Page experience. Long blocks add scroll depth, and above-grid blocks push the first row of products down. The 2024 removal test and the 2026 grid-size test both suggest listing pages are sensitive to this.
  • Source order and page weight. “Below the grid” is a visual position, and the text may sit anywhere in the HTML. On very heavy templates (large inline JSON, hundreds of tiles), copy at the end of the source is the first thing at risk, because Googlebot only fetches the first 2MB of an HTML file. Few stores hit that limit, but it is worth checking if yours is unusually large.

  Important distinction:

  An SEO split test tells you whether Google responds to the change. It does not tell you whether shoppers buy more. Above-grid copy can win on organic sessions and lose on revenue per session, so you may need both an SEO test (bucketed by page) and a conversion test (bucketed by user) before you decide.

  A decision framework before you test

  Condition	Lean towards a short intro above	Lean towards copy below	Lean towards little or none
  Query intent	Mixed or research-heavy categories where shoppers need to choose a type first (for example, “hiking boots for wide feet”)	Categories with real buying questions that only some shoppers have	Plainly transactional head terms where the products are the answer
  Product tile clarity	Product names are codes or brand names that do not describe the category	Tiles are clear but supporting detail (sizing, materials, compatibility) helps some buyers	Tiles, headings and filters already make the topic obvious
  Mobile first screen	The intro fits in two or three lines without pushing the first product row off screen	Any intro would push products below the first screen	The grid itself needs every pixel
  What you can actually say	You have one or two sentences that genuinely help selection	You have useful guidance that is longer than a shopper wants before products	The only available copy is generic or templated
  Category size	Thin categories with few products and little on-page context	Large categories with buying guides worth linking	Large categories where the grid carries the relevance

  How to test all three on one store

  Testing three arms needs more pages and traffic than a simple two-arm test, because each bucket gets a smaller share. SearchPilot says it usually recommends testing groups of pages that get “at least 1,000 sessions per day” for a reliable model, although large effects can show up on smaller groups with a longer run. A store with 40 category pages and modest traffic will not get a clean three-way answer, and should treat any result as directional.

  1. Hypothesis. Write one per arm. For example: “A two-line selection intro above the grid will increase organic sessions without reducing revenue per organic session.”
  2. Buckets. Randomly assign similar category pages into three groups, stratified by traffic and category type. Keep one bucket on the current layout as the control.
  3. Change only placement, or only length. Moving copy and rewriting it at the same time means you cannot tell which change mattered.
  4. Primary metric. Organic sessions or Search Console clicks per bucket against a forecast built from the control.
  5. Secondary metrics. Revenue per organic session, product click-through from the grid, add-to-cart rate, mobile scroll depth, and layout shift if the intro loads late.
  6. Duration and timing. Run for several weeks, avoid major sale periods, and annotate any core update that lands mid-test. A controlled design copes with updates better than a before-and-after comparison, but volatility still widens the uncertainty.

  Watch for this:

  SearchPilot’s test of moving product descriptions above the fold could not separate the effect of position from the effect of taking the text out of a JavaScript tab: “we can’t know for sure if it was because it was further up the page, we exposed it from the tab, or a combination of both factors.” Category copy tests fall into the same trap if the “above” variant also changes rendering, length or visibility.

  Hypothetical example:

  An outdoor store has 300 category pages. Each has a 250-word block below the grid, written years ago mostly to repeat the category name. The team splits 240 comparable pages into three buckets of 80: control (current block below), a two-line intro above the grid with the old block removed, and no copy. After six weeks, suppose the intro bucket is ahead on organic sessions, the no-copy bucket is level with control, and a separate user-bucketed conversion test shows the intro does not reduce add-to-cart. The decision would be to roll out short intros and drop the old blocks. If instead the no-copy bucket matched the intro bucket on traffic, the cheaper option wins and the writing budget moves to product data and filters.

  What results should change the decision

  • If “none” matches the other arms on both traffic and revenue, keep copy minimal and spend the effort on product data, filter pages and internal links.
  • If “above” wins on traffic but loses revenue per session, shorten it, move the longer guidance below, or collapse it on mobile while keeping it in the HTML.
  • If “below” loses to “none” on mobile only, as in the 2024 removal test, trim the mobile block or collapse it rather than assuming the text itself is the problem.
  • If removing copy costs traffic, as in the 2020 test, check which queries fell. That tells you what context the copy was providing, and you can often supply it more compactly.

  You do not need to write copy in a special way for AI features either. Google’s May 2026 guidance says artificial chunking and AI-specific rewording are not needed and that optimising for generative AI search is still SEO.

  Where I would start

  If this were my store, I would not pick a winner from anyone else’s test. I would default to a short, specific intro above the grid only where it helps shoppers choose and does not push the first product row off the mobile screen, put longer buying guidance below or on a linked guide, and delete generic blocks that exist only to repeat the category name. Then I would test the “none” arm on a subset, because it is the variant most stores never try, and it is the one that tells you whether your copy is earning its place at all.

  Need help with this on your own site?

  Niraj Raut works with businesses in Nepal, Australia, the UK, Europe and the US on technical SEO, ecommerce SEO, local SEO and AI search.

  See SEO services