TapCub is live — analytics, insights and live chat, free on one platform
Growth & Marketing

Ecommerce SEO tips: fix crawling and duplicates before content

Ecommerce SEO works in priority order: crawlable, distinct category and product pages first, content later. Filter and sort URLs, out-of-stock pages, a worked 10,000-to-1,200 URL example and how to measure orders.

Xu An Published Nov 17, 2026 16 min read Growth & Marketing
Sample data: a 20-category store with three filters and four sort orders. 8,800 of the 10,000 crawled URLs (88%) should be consolidated, blocked or removed; at most 1,200 are index candidates.Sample data

The most useful ecommerce SEO tips come in priority order: first make sure crawlers can reach and tell apart your category and product pages by handling filter, sort and out-of-stock URLs, then improve category and product content, and judge the result by orders from organic landing pages, not by rankings.

What you will learn

  • Fix crawling and duplicate URLs before content: in the sample, 8,800 of 10,000 crawled URLs (88%) should not compete for the index
  • Decide once per URL type: sort orders, empty and thin filter combinations, and the at most 1,200 states worth indexing
  • Keep out-of-stock pages that will return, and judge SEO by orders: 3,000 sessions at 3.0% beat 12,000 at 0.9% on efficiency, and nearly match on orders (90 vs 108)

Where to start: the priority order

Most ecommerce SEO advice is a flat list of twelve tips with equal weight. That hides the one thing that decides where your time goes: a store's biggest SEO problems are usually technical, and they come before content. A store with 10,000 near-identical filter pages and a hundred empty ones will not be rescued by better product descriptions.

Work in this order, and stop to measure after each step:

PriorityAreaWhat to check firstSignal that it needs work
1Crawlable and distinct pagesFilter and sort parameters, duplicate URLs, empty categories, out-of-stock pagesSearch Console shows thousands of crawled-but-not-indexed or duplicate URLs; crawlers spend most requests on parameter URLs
2Category pagesOne page per way shoppers search, with a title, intro and working internal linksTop-10 category pages with a low click-through rate; categories with no organic entries
3Product pagesUnique title and description, one URL per product, variants handled, structured dataManufacturer text copied across many stores; several URLs for one product
4ContentBuying guides and comparison pages for questions shoppers ask before buyingQuestions in chat and search queries with no page that answers them
5LinksInternal links to the pages that sell, then outside linksBest sellers reachable only through search or deep in the menu

Priority 1 comes first because it limits everything else. Google judges the importance of a page largely by how many internal links point at it, and a crawler that spends its visits on parameter URLs has fewer left for the pages that sell. The rest of this guide covers the priorities in that order, then how to tell whether any of it worked. It stays on structure and measurement. For turning organic visitors into orders once they arrive, see ecommerce CRO, and for a general plan when the team is small, SEO for startups.

Why filters and sort orders cause the most damage

Faceted navigation, the filters for colour, size, brand and price plus sort orders, can turn a hundred category pages into tens of thousands of URLs. Google's documentation on faceted navigation says plainly what that costs: crawlers can waste a lot of resources fetching large numbers of filter combinations before they realise those URLs are useless, which leaves less capacity for discovering new and useful pages.

Three kinds of URL cause the damage:

  • Sort and view parameters such as ?sort=price or ?view=48. They change the order, not the content, so they are duplicates of the base page.
  • Filter combinations such as ?color=black&size=9&brand=acme. Some match what people search for. Most do not, and many return no products at all.
  • Session and tracking parameters appended by the site or by campaigns. Google's URL structure guidance says not to link to temporary parameters like session IDs and tracking codes, and to use stable URLs.

Google lists the options for the URLs you do not want crawled, and they are not equal. A robots.txt disallow on the filter parameters is the stronger one, as long as category and product pages stay crawlable. A canonical tag pointing filtered URLs at the unfiltered version may gradually reduce crawling of the duplicates, but Google describes it as less reliable over time than robots.txt. Adding nofollow to links only helps if every link to that URL carries it. Filters placed after a # have no effect on crawling at all, because Google generally ignores fragments.

For the combinations you do want indexed, Google's advice is to use the standard & separator, keep filter order in the path consistent, and return a 404 status for combinations with no results rather than redirecting them to a generic page. For the reasoning behind parameter URLs, read dynamic URL. The table below turns this into a decision for each type of URL. Apply it to each type once, not to each URL.

What to do with each type of URL

The table gives a default treatment for each URL type. The right answer for a given store depends on search demand, so check the last column before applying it.

URL typeExampleTreatmentCheck before applying
Sort and view parameters?sort=priceCanonical to the base page, or disallow the parameter in robots.txtNone; they have no content of their own
Session and tracking parameters?sid=…, ?utm_source=…Do not link to them; keep stable URLs in internal linksWhich links generate them
Filter combination with demand?color=black&brand=acmeIndex it: self-referencing canonical, in the sitemap, & separator and consistent orderPeople search for it and it lists three or more products
Filter combination, thinOne or two productsnoindex, or canonical to the parent categoryWhether the parent already covers the query
Filter combination, emptyNo productsReturn 404; do not redirect to a generic pageWhether internal links still point to it
Filter placed after a ##color=blackDo not rely on it: Google generally ignores fragmentsIf it must be crawlable, use a real parameter or path
Variant with its own URL/shoe-blackCanonical to the main product URLSame URL form in links, sitemap and canonical

Treatments follow Google Search Central guidance on faceted navigation and ecommerce URL structure; which combinations deserve indexing depends on your own search demand.

Worked example: 10,000 URLs, 1,200 worth indexing

Sample data: a store has 20 categories. Each category has three filters: colour with 4 values, brand with 4 and size with 4. Each filter can also be left unset, so each category has (4 + 1) × (4 + 1) × (4 + 1) = 125 filter states, and 20 × 125 = 2,500 states in total, including the 20 plain category pages. Every state can be sorted 4 ways, so a crawl finds 2,500 × 4 = 10,000 URLs.

Apply the rules from the table:

  1. Sort variants. For each state, the default order is the real page and the other three sort orders are duplicates. That is 2,500 × 3 = 7,500 URLs with no content of their own.
  2. Empty results. Of the 2,500 states, 700 return no products. They should return a 404 or not be linkable.
  3. Thin results. A further 600 return only one or two products. Treat them as not worth indexing, with a canonical to the parent or a noindex.
  4. What is left. 2,500 − 700 − 600 = 1,200 states have three or more products, including the 20 category pages.

The partition adds up: 7,500 + 700 + 600 + 1,200 = 10,000. Only 1,200 URLs, 12% of what a crawler finds, are candidates for the index, and 8,800, or 88%, should be consolidated, blocked or removed. The 1,200 is a ceiling, not a target: of those states, keep only the ones that match something shoppers actually search, such as "black running shoes" or "acme size 9", and give those a self-referencing canonical and a place in the sitemap. The numbers are illustrative. The method is the point: classify every URL pattern by type, count each type and decide once.

app.tapcub.com

10,000 crawled URLs by what to do with them

10,000 URLs
Sort-order duplicates7,500
Empty results700
Too few products600
Candidates for the index (at most)1,200
Sample data
Sample data: a 20-category store with three filters and four sort orders. 8,800 of the 10,000 crawled URLs (88%) should be consolidated, blocked or removed; at most 1,200 are index candidates.

Category pages

Category pages usually carry more organic traffic than any single product, because they match broad searches like "women's running shoes". Three things matter most:

  • One page per search intent. If shoppers search for "trail running shoes" and "road running shoes" separately, they need separate pages with their own titles and intros, not a single page with a filter.
  • A title and a short intro that say what the page lists. Skip the long block of keyword text at the bottom; it helps nobody.
  • Crawlable links. Google recommends ordinary links in the page markup rather than click handlers, and linking menu to category to subcategory to product, so that every product you want indexed is reachable. Products found only through the site's own search box can be missed. If you cannot link everything, a sitemap or a Merchant Center feed helps.

Do not let two categories compete for the same query either. If "running shoes" and "jogging shoes" both have pages with nearly identical lists, see keyword cannibalization.

Product pages and variants

Product pages win on specifics: the exact product name, a model number, a size. Keep four rules:

  1. One URL per product. Google's ecommerce URL guidance says variants such as colour or size can have their own URLs, but if they do, the canonical should point at the main product URL. Use the same URL form in internal links, sitemaps and canonicals.
  2. Write your own title and description for the products that sell. Copy supplied by a manufacturer appears on every store that sells the item, so there is nothing to tell your page apart. Rewrite the top 50 sellers, not all 10,000.
  3. Add Product structured data with a price and availability so the page can be understood as a product and show current stock.
  4. Answer the question that stops the purchase near the buy button: delivery time, returns, sizing. That is both a conversion and a content fix, and it stops the question reaching your support inbox.

Out-of-stock and discontinued products

What you do with a page that cannot be bought depends on whether the product is coming back:

SituationTreatmentWhy
Out of stock for nowKeep the page live, mark it out of stock in the page and in structured dataGoogle's guidance on pausing a business says it is better to keep the page and mark it out of stock; shoppers and links keep their destination
Discontinued, a close replacement exists301 redirect to the replacementThe visitor and the link value both land somewhere useful
Discontinued, no replacementReturn 404 (or 410)Google treats 404 and 410 alike and removes the URL gradually
Seasonal, back next yearKeep the page and mark it out of stockKeeps its history and avoids rebuilding it
Empty category404, or noindex while emptyGoogle advises against indexing empty or low-value pages

A soft 404, a page that shows an error or "not found" message but returns a success status, can be flagged in Search Console. Do not use noindex on a page you will relist in a month, and do not redirect every discontinued product to the home page, which sends the visitor somewhere unrelated.

Content and links come fourth and fifth for a reason: they do not fix structure. Once the crawl is clean and category and product pages are sound, they are worth the time:

  • Buying guides that answer questions shoppers ask before they buy. Find them in your chat transcripts, support inbox and search queries, and link them to the category they help sell.
  • Internal links from guides and the home page to best sellers. This is cheap and quick.
  • Outside links from sites that matter in your niche, through guides, reviews and partnerships. This is the slowest and least controllable lever, so it should not hold up the rest.

Skip writing for keywords nobody searches. A page that no query, link or shopper question needs is one more URL to maintain.

Judge it by orders, not ranking

A higher ranking is a means, not the result. Two sample landing pages show why:

Landing pageOrganic sessionsConversion rateOrders
A: broad category12,0000.9%108
B: specific product3,0003.0%90

Page A has four times the traffic, but page B produces nearly as many orders from a quarter of the visits. If you celebrated A's rise from position 8 to position 5 and ignored B, you would have chosen the wrong page to work on. Together the two pages bring in 108 + 90 = 198 orders. Track at least three things for each landing page: organic sessions, conversion rate and orders. The wider set of store metrics is in ecommerce analytics.

When you ship a change, compare equal periods of four weeks before and after, and note the date on your charts. A parameter clean-up usually takes weeks to show in search, so judge it by the crawl and by the count of indexed duplicate URLs first, and by orders later.

Common mistakes

  • Noindexing every filter page. A few combinations match real searches and deserve to be indexed; sorting out which ones takes counting, not a blanket rule.
  • Believing canonical blocks crawling. A canonical consolidates signals, and Google says it reduces crawling of the duplicates less reliably than robots.txt does. If the crawl itself is the problem, block the parameters.
  • Deleting out-of-stock pages that will return. You lose the history and the links, and shoppers who find the page get an error.
  • Rewriting 10,000 descriptions before fixing 10,000 URLs. Fix the crawl first, then write for the top sellers.
  • Chasing rankings for terms that do not convert. A broad term may bring traffic with no orders. Count orders by landing page.
  • Counting bot visits as demand. Crawlers and scrapers can inflate sessions and distort conversion rate. Keep them out of the numbers you decide on.

Finding the cleanup priorities in TapCub

Take the 10,000-URL store above and use TapCub to check where it stands. TapCub measures what search and crawlers do; it is not a crawler or a site audit, and it does not change your robots rules, canonicals or noindex tags. Those stay on your store platform.

  1. Connect search data. In Monitoring › Search performance, connect Google Search Console or Bing Webmaster. You then see clicks, impressions, click-through rate and average position by page, query and country. Pages in the top 10 whose click-through rate is below the median are marked, because changing the title and description pays off first there. Search data runs 1 to 3 days late, and with no account connected, empty clicks mean no data, not zero traffic.
  2. See what crawlers fetch. Monitoring › Crawl health shows fetches, success rate, distinct pages, 404s and 5xx by engine, and if you set your sitemap URL it compares the sitemap with what was fetched. In the example, that comparison shows whether the 1,200 pages are being fetched. It needs server-side records, because the page script cannot provide a status code; see the crawl health help. Crawler 404s are separate from visitor 404s in the Web analytics › Pages 404 tab, which lists hits and the previous page. Fix paths that appear in both first.
  3. Know what the Pages report hides. When TapCub records a page it drops the query string, so every filter and sort variant appears under the same path in Web analytics › Pages. That keeps traffic readable, but it also means the report cannot show a parameter explosion. Use Search Console and crawl records for that. UTM tags are stored as source fields and appear in Web analytics › Sources.
  4. Measure orders by landing page. Define a goal on your order event with Track revenue on, and open Web analytics › Goals. The breakdown splits conversions by source, entry page, country or device: use the source split to see how organic search converts and the entry-page split to see which landing pages do. Humans and bots are counted separately by default, so recognised crawlers do not inflate sessions or dilute conversion rate. For multi-visit journeys, All-platform analytics › Behavior › Behavior analysis attribution can credit purchases to the landing page of earlier sessions within a 30-day window by default.

FAQ

What are the most important ecommerce SEO tips?

Fix crawling and duplicate URLs first, then category pages, then product pages, then content and links. Filter and sort parameters, empty categories and out-of-stock pages cause the biggest structural problems on most stores. Measure results by orders from organic landing pages rather than by ranking alone.

How do you do SEO for an ecommerce site?

Start with a clean crawl: consolidate or block parameter URLs, return 404 for empty combinations, and give each product one URL. Then write category pages for each way shoppers search, give top sellers their own titles and descriptions, add Product structured data, and track organic orders by landing page.

Should you noindex filtered pages?

Not all of them. Filter combinations that match real searches and have enough products can be indexed with a self-referencing canonical. Sort orders, empty results and thin combinations should be consolidated, blocked or removed. Count each type before deciding; a blanket noindex also removes the useful ones.

What should you do with out-of-stock product pages?

If the product is coming back, keep the page and mark it out of stock in the page and structured data. If it is discontinued and has a close replacement, 301 redirect to the replacement. If there is no replacement, return a 404 or 410. Do not redirect everything to the home page.

How long does ecommerce SEO take to work?

Technical clean-up shows first in crawl and index reports, often within weeks, while rankings and orders take longer. Compare equal periods before and after a change, such as four weeks each, and judge a parameter fix by the number of crawled duplicate URLs before judging it by sales.

Sources

Was this article helpful?

We use your answers to fix and expand this page.

Xu An

Customer Experience Consultant

Writes about live chat, visitor conversations, and self-service, sharing practical ways to make customer communication clearer and more effective.

See which pages search sends and which ones sell

Search performance by page and query, crawler fetches and 404s, and goal conversions by entry page, in one project.

See your data clearly. Find your growth.

Every click, backed by data. Install one line of code and see your first numbers in a minute.

No credit card · Free plans for analytics and chat · Cookieless analytics