A practitioner's guide to product page SEO: original copy, faceted navigation, honest schema, out-of-stock handling, and measuring revenue instead of impressions.
If you sell online, you already know the pattern. The store has a few thousand product URLs, a handful of them earn traffic, and the rest sit in Search Console with a trickle of impressions and no clicks. Someone suggests optimising the product pages, which in practice means rewriting title tags and working the keyword in a few more times. Nothing moves.
The reasons product pages rank or don't are fairly well understood, and most of them are structural rather than cosmetic. What follows is the working list: what matters, why it matters, and how the work is actually done. There are no percentages in this article, because anyone quoting you a number before they have crawled your store is guessing.
The manufacturer's description is not your content
If you take the supplier feed and publish it as-is, your description exists word-for-word on every other store selling that SKU, plus the manufacturer's own site, plus the marketplaces. Google has to pick one page to show for that product, and when the copy is identical the decision comes down to everything else: domain authority, links, engagement, brand. That is a contest a mid-sized store loses by default.
Original product copy is the single biggest differentiator available to you, and it is the one most stores skip because it is slow. Specification tables can and should stay factual and repeatable — nobody expects a unique way to state a voltage rating. It is the prose around them that has to be yours: what the product is actually good for, what it is not good for, sizing and fit as your customers report it, compatibility, what is in the box, how it compares to the two products next to it in your own catalogue.
You are not going to rewrite four thousand SKUs, and you shouldn't try. Rank the catalogue by revenue and by impressions, rewrite the top slice by hand, and use a structured, variable-driven template for the long tail so at least the shape of each page is not identical. Then work down the list as capacity allows.
Category pages carry the head terms. Product pages chase the long tail.
Most stores have this backwards, and it costs them the commercial keywords entirely.
Somebody searching men's trail running shoes is shopping a category, not a SKU, and the result that satisfies them is a well-built category page with a curated set of products. Somebody searching a full model name with a colourway and a width fitting is shopping a specific product. Both are valuable. They are not the same page, and pointing your best links and best copy at individual product URLs while leaving category pages as bare grids of thumbnails is how a store ends up ranking for nothing with volume.
Practically: category and collection pages deserve real introductory copy, sensible buying guidance, and internal links out to the subcategories and products that matter. Product pages should be built for specificity — exact model, variant, use case, compatibility. Getting this split right is usually the first structural fix in any eCommerce SEO engagement, and it often changes more than a year of on-page tweaking.
Faceted navigation is where crawl budget goes to die
This is the most common technical failure in eCommerce, and it is almost always invisible from the front end.
Filters multiply. Colour times size times brand times price band times sort order produces an enormous number of URL combinations, and if each of those combinations is reachable through a normal crawlable link, Googlebot will happily spend its time on them instead of on your new products. You end up with a crawl log full of ?colour=blue&size=10&sort=price_desc&page=4 and a set of genuinely important URLs that haven't been crawled in months.
The fix is a decision, not a plugin. Work out which facet combinations have real search demand — usually brand plus category, sometimes colour or size plus category — and promote those to proper indexable pages with clean, static, lowercase URLs and their own copy. Everything else needs to be kept out of the crawl. In rough order of effectiveness:
- Don't create the link. The most reliable way to stop a URL being crawled is not to expose a crawlable a href to it. Render low-value filter states without linkable hrefs.
- Disallow the parameter pattern in robots.txt once you are certain nothing valuable lives behind it. This stops crawling, which is the actual resource you are protecting.
- Use noindex when you need the page usable and crawlable but out of the index. Note the trade-off: a noindexed page still has to be crawled to be seen, so it saves index bloat, not crawl budget.
- Canonicalise filtered variants back to the clean category URL. Treat this as a hint, not an instruction — Google can and does ignore canonicals it disagrees with.
Two traps worth naming. First, a URL disallowed in robots.txt can never be crawled, so a noindex tag on it will never be read — pick one mechanism per URL pattern, not both. Second, sort orders, session identifiers, and tracking parameters should never be crawlable at all. Check the Crawl Stats report and, if you can get them, the server logs; they will tell you within minutes where the crawler is actually spending its time. This is the bulk of what technical SEO work on a store consists of.
Product schema, done honestly
Product structured data is worth implementing: name, image, description, SKU and GTIN or MPN, brand, and an offers block with price, currency, availability, and URL. The single rule is that the markup must reflect what is actually on the page and actually true. If the markup says in-stock and the page says sold out, or the price in the JSON-LD is stale, you are inviting an eligibility problem and disappointing the person who clicks.
Reviews are where stores get themselves into real trouble. AggregateRating markup is only legitimate if the ratings were genuinely left by customers, are visible on the page, and belong to that product.
Marking up review ratings you did not genuinely receive — invented, self-issued, or imported from somewhere else and presented as your own — is a manual action risk. It can cost you rich results across the entire domain, and reinstatement is slow. Do not do it, and do not let anyone do it for you.
If a product has no reviews, ship it without review markup. A page showing five stars and zero reviews is a signal to a manual reviewer, not a shortcut. The sustainable version of this is boring: collect real reviews with post-purchase emails, publish them including the mediocre ones, and let the markup follow reality. Feed-driven markup is far easier to keep honest than hand-edited templates.
Out of stock, discontinued, and the temptation to delete
Deleting product URLs at scale is one of the most expensive habits in eCommerce. Those pages have accumulated links, internal signals, and search history. A mass 404 event throws that away and gives you nothing back. Handle each case on its merits:
- Temporarily out of stock, coming back: keep the URL live and returning a 200. Set availability to OutOfStock in the markup, say so honestly on the page, and offer a restock notification plus real alternatives.
- Discontinued with a clear successor: 301 to the successor product. The intent carries over cleanly.
- Discontinued with no successor: the honest option is usually to keep the page live, state plainly that it is no longer available, and link to the closest alternatives and the parent category. Redirecting hundreds of dead products into one category page tends to get treated as a soft 404 anyway. Remove them from navigation and sitemaps over time.
- Genuinely gone, no equivalent, no traffic, no links: a 410 is fine. It is honest and it cleans up.
Orphaned products never rank
If the only route to a product is the XML sitemap or the internal search box, it may get crawled, but it will never pick up the internal link signals it needs to compete. Every product should be reachable through indexable category pages within a few clicks of the homepage.
The audit is straightforward: crawl the site while ignoring the sitemap, then diff the crawled product URLs against the full product list from your database or platform export. Anything in the catalogue and not in the crawl is orphaned. The usual culprits are client-side pagination on category pages — where Google only ever sees page one — and related-product modules that render in JavaScript with no real links behind them. The fix is server-rendered anchors and genuine paginated URLs. On hosted platforms this is a theme-level job rather than an app-level one, which is why Shopify SEO work so often ends up inside the Liquid templates.
Images and alt text
Alt text exists to describe the product to someone who cannot see the image. Brooks Ghost trail shoe, navy, side profile is useful. A string of keywords and your city name is not, and it carries risk with no upside. Write the description, move on.
The rest of image work is mechanical: descriptive filenames, WebP or AVIF with sensible fallbacks, correctly sized responsive variants instead of a 3000px master scaled down in the browser, lazy loading below the fold — and never on the main product image, which is almost always the Largest Contentful Paint element and should be eagerly loaded and preloaded.
Speed, on template-heavy stores
Product templates accumulate weight the way garages accumulate boxes. A review widget, a chat bubble, an upsell app, three analytics tags, a personalisation script, an A/B testing tool. Every one of them is JavaScript running on your highest-intent page.
The audit is an inventory: list every third-party script on the product template, find who owns it, and delete the ones nobody can justify. Defer what survives. Fix the LCP image. Strip unused theme CSS. Speed is a modest ranking input but a substantial conversion input, which is why this work usually sits alongside CRO rather than apart from it.
Measure what pays
Impressions and average position are the easiest numbers to make go up and the least connected to your bank account. Report on organic revenue by landing page, add-to-cart rate for organic sessions, and the ratio of published products to products that earned at least one organic click in the last 28 days. That last metric is the honest scoreboard for product-page work: if it climbs, the work is landing. If it doesn't, something structural is still broken.
A sane order of operations
- Fix crawl waste first — facets, parameters, sort orders. Everything downstream depends on Google being able to reach your real pages.
- Fix orphans and internal linking so products actually receive signals.
- Build category pages that can hold the commercial head terms.
- Rewrite copy on the highest-revenue and highest-impression products, then work down.
- Make the schema true, including the parts that are unflattering.
- Then speed, then conversion.
None of this is exotic and none of it is fast. If an agency has told you otherwise, that is the part worth being suspicious of. If you want someone to run the crawl, read the logs, and tell you which of these actually apply to your store before anyone touches a template, that is where we would start too.
Need Professional Digital Marketing Services?
Our expert team is ready to help grow your business online.