The difference between on-page SEO on a blog and on a store is arithmetic. A blog has forty pages and you can hand-edit every one. A store has four thousand, and hand-editing is a plan that never finishes. Everything worth doing in on page SEO ecommerce gets done once, in a template, and inherited by the whole catalogue — with hand-crafted exceptions reserved for the pages that pay for them.
That constraint shapes every decision below. Ask of each tactic: can this be generated reliably from data I already have, and will it still read well on the ten-thousandth page?
Decide What Each Page Type Is Competing For
Before touching a title tag, settle the mapping. Category pages carry the commercial head terms — “office chairs,” “merino base layers,” “cordless drills” — and typically hold ten to fifty times the monthly demand of any single product inside them. Product pages win brand-plus-model queries and the long tail of specific intent: “Herman Miller Aeron size B remastered.”
Confusing these is the most expensive on-page mistake in ecommerce. Optimising a product page for a head category term means competing against category pages with far more topical breadth and internal links, and losing. Optimising a category page for a specific model means burying the one page that could have converted that searcher immediately.
Map it explicitly: one primary query per URL, documented in a spreadsheet, revisited quarterly. Keyword research that returns up to 150 ideas per seed with volume, difficulty and CPC makes this a two-hour job instead of a two-week one, and it exposes the categories you never built.
Title and Meta Templates Worth Deploying
Titles are the highest-leverage template field because they influence both relevance and click-through. Build them from fields you actually maintain.
- Product:
{Brand} {Model} {Key Attribute} | {Store}. Keep it under roughly 60 characters so it survives truncation; drop the store name first when it overflows. - Category:
{Category} — {Qualifier} | {Store}, where the qualifier is a real modifier people search (“for men,” “waterproof,” “with free delivery”), not a stuffed synonym. - Facet landing pages: only where you have deliberately chosen to index them, built from the facet values so no two collide.
Meta descriptions will often be rewritten by Google, and that is fine — a good one still earns clicks when it survives. Template in the things a shopper wants to know before clicking: price range, stock status, delivery promise, return window. “Shop our wide range of quality products” earns nothing.
The failure mode to hunt for is duplication. Run a crawl and sort by title; near-identical titles across dozens of URLs mean a template variable is empty or a facet is leaking into the index. Crawls that hand back the actual title text and URL rather than a count are what make that fix a ten-minute job.
Write Category Copy That Helps Someone Choose
Category pages are where most stores put a block of keyword-padded text under the product grid and call it content. Shoppers do not read it and Google is not impressed by it. What works is buying guidance: how to pick a size, what materials cost more and why, what the difference is between the entry model and the premium one, which use case suits which sub-range.
Structure it for skimming. A short intro of two or three sentences above the grid to confirm the shopper is in the right place, then the substantive guidance below the products, broken into subheadings with a short FAQ addressing the questions your support team answers every day. Two hundred honest words beat eight hundred padded ones.
One practical constraint: category copy has to survive merchandising changes. Do not reference specific product counts or prices that go stale in a week.
Product Descriptions Are Where Duplication Kills You
If you sell products that other retailers also sell, and you pasted the manufacturer’s description, your page is one of hundreds carrying byte-identical text. Google chooses one canonical answer to that duplication and it is almost never the smallest merchant. This is the single most common thin-content trap in ecommerce, and it is entirely avoidable.
Prioritise instead of trying to rewrite everything at once. Rank SKUs by revenue and by search demand, then:
- Top decile: full original treatment — what ownership is actually like, who it suits, what it does poorly, sizing quirks, honest comparison to the obvious alternative.
- Middle: structured specification tables plus three or four original sentences covering fit, use case and the one thing customers ask about most.
- Long tail: specs, images, unique title and H1, and at minimum a single original line. Never a copy-paste block.
Volume is the real bottleneck, which is why an AI writer earns its place here — SEO Rocket’s writer runs a proven template behind hard validation gates and supports brand voice plus an uploaded brand guide, so catalogue-scale copy still sounds like your store instead of a datasheet. Feed it your genuine product knowledge; feed it nothing and you get boilerplate with a different accent.
Headings, Images and the Bits Themes Break
Every page needs exactly one H1, matching the product or category name. Themes break this constantly — a promo slider or a logo wrapped in a heading tag gives you two or three. Check the rendered DOM, not the source file, because client-side rendering changes the answer.
Subheadings should describe real sections (Specifications, Sizing, Care, Delivery and Returns), not repeat the product name in slightly different word orders. Alt text should describe what is in the image — “navy wool overcoat, front view, on model” — which serves accessibility and image search in the same stroke. In visual categories like fashion, furniture and home goods, image results drive a meaningful share of discovery, and filename plus alt text is most of what you control.
Breadcrumbs deserve a mention here because they are on-page and structural at once: real anchor links, marked up with BreadcrumbList, reinforcing the hierarchy and passing internal link equity upward for free.
Structured Data as an On-Page Deliverable
Treat schema as part of the template, not a separate project. Product markup with name, image, sku, brand, description and a complete offers block drives price and availability in the search results, and availability that lags the page by a day is worse than no markup at all.
Mark up genuine customer reviews of the product only. Self-serving review markup — a business marking up praise of itself outside real product reviews — is not eligible for rich results and risks having your snippets suppressed entirely. The stars are not worth that trade.
Internal Links Belong in the On-Page Plan
On-page work stops at the page boundary for most people, which is why so many stores have perfectly optimised pages nobody can reach. Related products, cross-category modules and editorial links into commercial pages are all on-page elements, and they should be curated rather than randomised.
Two rules that hold up: keep related-product modules stable across page loads so crawlers see a consistent graph, and cap them at a useful number rather than dumping forty links onto every product page. Descriptive anchor text matters — “waterproof hiking boots” tells a crawler something, “click here” does not.
Verify at Scale, Then Measure Honestly
The only reliable way to confirm template-level on page SEO ecommerce work is a full crawl after deployment. Spot-checking three pages proves nothing when the bug only affects products with empty brand fields. Look for duplicate titles, missing H1s, thin pages under a word-count floor, orphaned URLs and broken canonicals — with the actual evidence attached so you can trace each one back to the template rule that caused it.
Then be patient with the numbers. Two or three positions of daily movement is ordinary noise across any catalogue, and third-party position and volume data are modeled estimates from periodic crawls. Segment Search Console by page type, watch impressions and click-through for category and product pages separately, and give template changes six weeks before deciding. SEO Rocket runs the crawl, the keyword mapping, the writer and the tracking in one place at a flat US$50 a month; the sequencing above is what makes any of them worth running.
