Internal linking for ecommerce, done deliberately

Before planning anything, it is worth separating where the links on your store come from, because they are not equivalent and most internal linking advice treats them as if they were.

Navigation menus. Configured in the admin, rendered from the linklist object, and capped at three nested levels — the documentation is explicit that a menu supports "a maximum of 3 levels". Every page carries them, which is exactly why they carry so little weight. A link that appears on all 4,000 URLs on your site distinguishes nothing from anything.

Collection membership. A product belongs to collections; the collection grid links to the product. This is the layer that makes your catalogue reachable at all, and the main thing to get right is that no product is in exactly zero collections, which happens more often than people expect after a bulk import.

Recommendation blocks. "You may also like", "Pairs well with". Merchandising, discussed below, and not the linking layer you think it is.

Links somebody wrote into a sentence. A paragraph on a hub collection pointing at a narrower one. A blog post pointing at the collection it is really about. These are the only links on the store that reflect a judgement about what matters, and they are the ones that are usually missing.

This is the correction worth making first, because it changes what you expect from the section your theme already ships.

Shopify exposes product recommendations through an Ajax endpoint: GET /{locale}/recommendations/products.json for data, or /recommendations/products with a section_id to render a section server-side. It takes a product_id, an optional limit that the documentation says "can range from 1 to 10, and the default is 10", and an optional intent of related or complementary — related uses relationship-based strategies, complementary uses co-purchase data and is the one you want under a buy button. The Search & Discovery app can override what comes back, which is the supported way to force a pairing the algorithm has not worked out yet.

The catch is the word Ajax. The block is fetched after the page loads, so the links inside it are not in the HTML the server returned. Treat that section as what it is — a conversion device that raises average order value — and not as the mechanism by which your accessories collection gains authority. If you want a crawlable link to that collection, put one in the page copy.

Which is not an argument for removing it. Complementary recommendations under a buy button earn their place on commercial grounds alone. It is an argument for not counting them twice.

Here is the whole method, and it fits in a spreadsheet.

Write down the twenty pages you want to rank — almost always collections, for the reasons in our piece on collection pages: category-level phrases carry the demand and a collection outlives the products inside it. Next to each, the phrase you want it to rank for. Then, for each destination, name three pages that already have authority and a plausible reason to mention it: the parent collection, a related collection, a blog post on the subject.

Now write the sentences. Not "click here", not a bare product name — the phrase the destination is targeting, inside a sentence that a human reading the page would find useful. If you cannot write a sentence on the source page that naturally mentions the destination, that is a signal the link does not belong there, and forcing it in is how a site ends up with a footer full of keyword links that fool nobody.

Twenty destinations times three sources is sixty links. That is an afternoon, it is the whole job for most stores under a few thousand products, and it is measurably more than an app that automatically links every occurrence of every product name will ever do for you.

The structural layer is not where the wins are, but it is where the faults are.

Breadcrumbs. Worth having, mainly because they make the hierarchy legible to a shopper who landed from search and to a crawler working out what sits under what. Mark them up as BreadcrumbList and — the part that gets missed — make sure the visible trail and the markup agree. A trail that stops at the parent while the schema includes the current page is a disagreement somebody eventually notices.

Orphans. Pages reachable only from the sitemap, or only from a filter control. Run your catalogue against your collections and find the products in nothing; check whether the facet collections you built last quarter are linked from anywhere other than the menu you removed them from.

Depth. If a product needs five clicks from the homepage, the pages in between are doing a bad job of linking. Three is a reasonable target for anything you care about, and the fix is usually a card row on a hub page, not another level of nav — the three-level cap means the nav cannot solve it anyway.

None of this is glamorous and all of it is cheaper than the alternative, which is discovering in month nine that a third of your catalogue was never linked from a page a crawler visits.

Teams rarely disagree with any of the above. They stall on production. Sixty in-copy links means editing sixty pages, and on a theme where a collection page is a title and a grid, adding a paragraph with two links in it is a developer ticket. Sixty developer tickets do not get raised.

That is a tooling problem rather than an SEO problem, and it has a direct fix: composable blocks a merchandiser arranges without touching code, so an intro paragraph with links, or a row of cards pointing at four sibling collections, costs ten minutes rather than a sprint. That is what a [section library](/services/build/shopify-section-library) is for, and internal linking is one of the clearest returns on building one.

The honest version of the advice: if your team cannot edit collection copy today, fix that first. The link plan is worthless while it stays in a spreadsheet.

• Shopify section libraries for content-heavy brands
• Collection pages are your best SEO asset
• Faceted navigation without wrecking your crawl budget