E-commerce Search Visibility Disputes

Why retail disputes look different

A retail site is not a brochure site with more pages. Its architecture is generated: category pages from a taxonomy, product pages from a catalog, and filter combinations on demand from whatever the shopper clicks.

Three consequences follow. A defect is rarely confined to one page — it propagates across a template and so across thousands of URLs at once, which makes the loss large and the mechanism specific. The failure modes are technical and leave records, so they are testable in a way a content-quality dispute never is. And the retailer's platform limits both what the vendor could have done and what evidence now exists.

The consistent shape is that a retail visibility case is a causation case fought at URL level. Site-level totals are almost useless. The question is which URLs, carrying which defect, lost which traffic on which dates — and the answer is normally reachable from records the parties already hold.

Replatforming, and the sentence in Google's guidance that both sides quote

Migration disputes are the most common retail search matter, because a replatform touches every URL at once and the failure modes are documented.

Google's guidance on moving a site with URL changes, last updated 17 June 2026, is specific enough to serve as a checklist. It recommends HTTP permanent redirects — 301 and 308. Googlebot follows redirect chains up to ten hops, but Google advises keeping the chain ideally no more than 3 and fewer than 5. The Change of Address tool applies to domain migrations only, and must be submitted for all subdomains and both www and non-www variants. Redirects should be kept for as long as possible, generally at least 1 year.

Then comes the sentence that sets the tone. Google states that a small to medium site can take a few weeks for most pages to move, and that you may experience ranking fluctuations while Google recrawls and reindexes your site. That is the defense's best line and the claimant's worst: a post-migration decline proves nothing on its own.

Establishing that a migration failed requires identifying the defect — a missing redirect, a chain, a loop, a 302 where a 301 was required, a canonical tag pointing back at the retired host — and tying URL-level traffic loss to the URLs that carried it. Google's site-move guidance is the benchmark; the redirect map is the proof.

The platform constrains both the work and the evidence

Before any opinion about what the vendor should have done, establish what the platform allowed. Platforms differ substantially in the control they give over URL structure, robots directives, redirects, and log access, and the constraints must be pinned to the platform and version in the matter.

Shopify is the clearest illustration, observable by inspection of any store. Path segments are fixed: products live under /products/, collections under /collections/, static pages under /pages/, blog articles under /blogs/. Hierarchical category paths are not available. A product in three collections is reachable at several URLs, which makes canonical handling a live issue. Sitemaps generate automatically and cannot be replaced. Most consequentially for litigation, raw server access logs are not available to the merchant.

That changes the achievable scope of an opinion. Without log access, the best source for proving what Googlebot actually requested does not exist, and the opinion must be rebuilt from URL Inspection results, crawl exports, and archived captures.

Version support dates matter for a different reason. Adobe Commerce publishes a three-year standard support window from each version's general availability date: 2.4.5 support ended 12 August 2025, 2.4.6 on 11 August 2026, and 2.4.7 runs to 31 May 2027. Where the defense is that a migration was unavoidable, the published end-of-support date either corroborates that or it does not.

Faceted navigation and crawl waste

Faceted navigation is the filter interface on a category page — color, size, price band, brand — where each selection generates a distinct URL. It is the most productive area of technical opinion in retail matters, and Google states the harm in its own words.

Because faceted URLs look novel, and a crawler cannot tell whether one is useful without fetching it, Google says crawlers will typically access a very large number of faceted navigation URLs before determining they are useless. The consequence Google names is direct: crawlers then have less time for new, useful URLs.

Google's documented remedies are specific: disallow patterns in robots.txt that block a filter parameter while still permitting item pages; a rel="canonical" from filtered views to the unfiltered version, which Google says decreases crawl volume of the non-canonical versions over time; and, where facets must be indexable, a consistent filter order and an HTTP 404 when a combination returns no results.

Google's crawl budget guidance, updated 22 July 2026, gives the thresholds at which this begins to matter: large sites of a million or more unique pages, and medium or larger sites of ten thousand or more with rapidly changing content. It also names the wastes — duplicate content, soft 404s, long redirect chains — and notes that noindex is not a crawl control, because Google still requests those pages.

This is the strongest technical opinion available in retail matters because it is measurable and requires no assumption about ranking algorithms. Count Googlebot requests by URL pattern in the logs, show the proportion consumed by parameterized facet URLs, and show the crawl rate on product and category URLs before and after.

Product and category indexation

The mechanics are simple enough to explain to a jury and specific enough to prove.

Google discovers products by crawling links and by reading sitemaps. Its e-commerce guidance is explicit that if category pages do not link directly to all products, Googlebot might not find every product by crawling alone, and it recommends ordinary <a href> navigation rather than JavaScript click handlers. Google also treats internal linking as an importance signal: the more links a page receives from within a site, the higher its importance relative to other pages on that site.

The testable failure modes follow directly:

• Infinite-scroll category pages with no crawlable pagination
• Products reachable only through a JavaScript filter, with no static path
• Category pages set to noindex during a redesign and never reverted
• Canonical tags on product pages pointing at a category page
• A submitted sitemap listing URLs that return 404
• A robots.txt Disallow rule added for a staging deploy and carried to production

Each is binary. Either the directive was present on a given date or it was not, and archived captures, deployment records, and crawl exports settle it. That is why indexation opinions survive scrutiny better than ranking opinions.

Structured data, rich results, and one rare documentary artifact

Structured data is machine-readable markup describing what a page is about — for a retailer, the product name, price, availability, and review information. It qualifies pages for enhanced result formats.

Google distinguishes two product experiences. Product snippets apply to pages where a purchase cannot be completed directly, and carry more options for review information. Merchant listings apply to pages where a customer can buy, and carry more options for detailed information such as sizing, shipping, and return policy. Supplying the required merchant-listing properties also makes a page eligible for product snippets. Availability uses schema.org values including InStock, OutOfStock, and Discontinued.

Two facts matter disproportionately in litigation. The first is that a structured data issue is a named manual action type — markup violating structured data guidelines. That makes markup a rare corner of technical SEO where a defect produces a documented, engine-generated notice with a date on it.

The second is that result types get discontinued. Google's data anomalies record shows FAQ rich results permanently discontinued as of 7 May 2026, so a retainer promising FAQ schema for rich results is now promising something that no longer exists.

Out-of-stock and discontinued inventory

No area of retail search produces more confident and less supportable testimony, so it is worth being precise about what is documented.

Google does not publish, in any documentation I have retrieved, an instruction on which HTTP status code to return for an out-of-stock or discontinued product. An expert who testifies that Google requires particular handling for retired product URLs should be asked for the page and its date.

What is documented is adjacent but useful. Google's crawl budget guidance states that a 404 is a strong signal not to crawl a URL again, and that 503 and 429 reduce crawl allocation. Google's removals documentation states that permanent removal requires a 404 or 410 status, password protection, or a noindex tag, and that robots.txt is not a blocking mechanism. And the product markup vocabulary supports OutOfStock and Discontinued, so a retailer can express the state without removing the page.

The trade-off is real. Keep discontinued URLs live and the site accumulates thin pages. Return 404 and the accumulated links and rankings go with them. Redirect them to a category or successor product and risk soft-404 treatment. Each is defensible. What is testable is not which approach was right, but whether it was implemented consistently and with what consequence.

Marketplace listings against the owned channel

The recurring commercial version of this dispute: a brand sells on its own site and on marketplaces, and the marketplace listing outranks the brand's own product page for its own product query. The claim is that an agency, a distributor, or a reseller degraded the owned channel — by syndicating identical product copy to the marketplace, by supplying it a richer product data set, or by permitting unauthorized resellers to list.

What is technically establishable is a decent list. Which URL ranks for the query, at which date and in which locale. Whether the product descriptions on the two properties are identical, and which published first. Whether the owned pages carry complete merchant-listing structured data while the marketplace listing does. Whether they are indexed at all.

What is not establishable, and must not be asserted, is that Google favors marketplaces. That is a conclusion about ranking systems that no available evidence supports, and I am aware of no engine statement about how marketplace and brand-owned pages rank against one another. It is the kind of assertion that draws a motion to exclude and takes the rest of the opinion down with it.

The disciplined formulation is comparative and factual: what each property published, which appeared for the query on these dates, and what changed on the owned site.

Frequently Asked Questions

Published 2026-08-11