What a platform decides for SEO
Every content management system and store platform makes choices on the site's behalf. Some fix the shape of URLs, adding folders or identifiers the site cannot remove. Some generate pages nobody asked for, such as tag archives, attachment pages or filtered views, each with its own URL. Some render pages in the browser with JavaScript, others on the server. Some let every template carry its own robots rules and canonical; others offer one switch for the whole site. A team that chooses on editorial features, price and design alone inherits all of these choices and discovers them in the first audit.
The answer is to treat SEO capabilities as requirements, alongside cost, integrations and editorial workflow, and to test them on a trial install rather than trusting a feature list. The general method for evaluating tools, from requirements to trials to a decision record, is in the guide to choosing and evaluating tools. This guide supplies the SEO requirements for it. Comparisons of particular platforms for marketing as a whole, such as the comparison of Shopify and WooCommerce, are a useful second step once the requirements are written.
Control over URLs and redirects
Google's SEO Starter Guide recommends descriptive URLs that contain words useful to people and grouping topically similar pages in directories. A platform that fixes URL patterns can make both impossible. Require:
- Editable URL slugs per page, and editable or removable fixed segments for categories, products, articles and collections.
- A choice of structure, so that sections can live in the folders the site architecture calls for.
- One form per URL, with the platform enforcing case, trailing slash and host and redirecting the others.
- Server-side permanent redirects that editors can manage, in bulk as well as one at a time. Google's redirects documentation recommends permanent server-side redirects when a URL changes and describes meta refresh and JavaScript redirects as fallbacks for when server-side redirects are not possible, with JavaScript the least reliable because it depends on renderingThe step in which a search engine runs a page's HTML, CSS and JavaScript to build the content a browser would display before indexing it..
- Automatic redirects when a slug changes, which an editor can see and override.
- Pattern-based redirects for migrations, so thousands of old URLs can be mapped by rule.
Redirect capacity matters most at the moment of moving onto the platform. The site migrations guide covers the move itself; the evaluation only has to confirm that the platform can hold the redirect map the move will need.
Indexing controls and canonicals
Search engines need to be told which pages to index and which URL represents each piece of content. The platform must let the site say both, per page and per template:
- A self-referencing Canonical TagTells search engines the preferred URL. on every indexable page, built from the clean URL, and the ability to override it per page.
- Robots meta rules per template and per page, so tag archives, internal search and thank-you pages can be set to noindex without touching articles or products. A single site-wide switch is not enough.
- Response headers for non-HTML files, so documents can carry indexing rules too.
- An editable robots.txt, version-controlled or at least logged.
- An XML SitemapA file listing the URLs a site wants search engines to discover and crawl. generated automatically, listing only canonical, indexable URLs, and excluding pages set to noindex.
- Control over generated pages: whether tag, author, date, attachment and filtered views exist at all, and how their parameters behave.
- Environment separation, so a staging or preview site can be protected without settings that might be deployed to production.
The test is simple: on the trial install, set a template to noindex, change a canonical, add a tag and a filter, then crawl the site and read what each URL returns.
Templates, metadata and structured data
Titles, descriptions and markup should come from data, with editors able to override them. Google's Starter Guide asks for titles and descriptions that are unique and describe the page; on a site of thousands of pages that happens only if the template builds them from fields. Require:
- Editable title and meta description per page, with template patterns and fallbacks for empty fields.
- Control of headings, so the main heading is the page's subject and not the site name or a logo.
- Alt text fields on images that templates actually output.
- Structured data generated from fields. Google's introduction to structured data describes it as a standardised format for providing information about a page and recommends JSON-LD as the format that is easiest to implement and maintain. A platform that outputs Schema MarkupStructured data vocabulary. from product, article or location fields, and lets developers extend it, keeps markup consistent with the visible content.
- Custom templates and fields, so new page types can be built without workarounds.
- Internal linking tools: breadcrumbs, related content and navigation that produce plain links.
Rendering and page speed
Google's JavaScript basics describe crawling, rendering and indexing as separate phases and recommend server-side rendering or pre-rendering because it makes sites faster for users and crawlers and because not every crawler runs JavaScript. It also asks for real URLs for each view using the History API rather than fragments, and for meaningful status codesThe numeric code a web server returns with every response, saying whether the request succeeded, was redirected or failed and why., including not-found responses for missing pages. Platforms differ sharply here. Require:
- Content and links in the server response for every indexable template, or a supported server-rendering mode for a script-heavy front end.
- Real status codes: a missing product or article returns not found, not a success page saying it is gone.
- Performance on real templates. Test the category, product and article templates of the trial install with realistic content and the scripts the site will add, and compare the results in the terms the Core Web Vitals guide uses. A demonstration theme with no images and no tags says little.
- Control over scripts and images: the ability to remove platform scripts that are not needed, set image sizes and formats, and choose what loads first.
The JavaScript SEO guide explains the failure modes to test for on platforms that render in the browser.
International and multi-site features
Sites serving more than one language or country need the platform to express that. Google's documentation on localised versions describes three ways to declare them: link elements in the HTML head, HTTP headers and the sitemap. It requires each version to list itself and every other version. Require:
- hreflangAnnotation for language/region versions. support by at least one of those methods, generated from the platform's own record of which pages are translations of which, not maintained by hand.
- The URL structure the international plan calls for, whether separate domains, subdomains or folders.
- Per-locale metadata, slugs and structured data, not a single field translated in place.
- No forced redirects by location or language, or the ability to turn them off.
- Shared or separate content across sites, depending on whether regional editions share pages.
The hreflang guide covers implementation; the evaluation checks that the platform can produce it without a spreadsheet.
Running the evaluation
Write the requirements as a scorecard before looking at demonstrations, mark each as a must-have or a should-have for this site, and test every one on a trial install with real content. Ask the vendor or the implementation partner to show the capability working, not to confirm it exists.
| Requirement | Priority | How to test on a trial install |
|---|---|---|
| Editable URLs for every page type | Must | Change a category and a product slug; check the URL and the redirect |
| Server-side permanent redirects, single and bulk | Must | Import a redirect file; request old URLs and read the status |
| Noindex per template and per page | Must | Set the tag archive to noindex; read the head and headers |
| Canonical per page, with override | Must | Inspect canonicals on a page, a filtered view and a parameter URL |
| Automatic sitemap of indexable URLs only | Must | Fetch the sitemap after setting a page to noindex |
| Titles and descriptions from fields, with overrides | Must | Edit fields; check output and fallback when empty |
| Content and links in the server response | Must | Fetch a template with scripts disabled |
| Structured data from fields, extendable | Should | Validate markup on product and article templates |
| Performance on real templates | Should | Lab tests on real content and the scripts to be added |
| International annotations from translation records | Should, or must for multilingual sites | Inspect the hreflang output for a translated page |
| Control over generated pages | Should | Create tags and filters; crawl and list the new URLs |
Record each result with evidence: a screenshot, a fetched response or a crawl export. A capability that needs a plugin, an app or custom development is a pass with a cost, and the cost belongs in the decision record. Once a platform is chosen, the SEO requirements tutorial turns the same capabilities into specifications for each template the build will create.
Cobwhisk Bakeware scores two platforms
Cobwhisk Bakeware, an invented bakeware brand with about 400 products and a recipe library of 250 articles, shortlists two store platforms, called here Platform A and Platform B. The team writes 11 requirements, 7 of them must-haves, and spends two days on trial installs with 50 real products and 20 articles. Platform A passes all 7 must-haves, 2 of them through an add-on with a monthly fee. Platform B fails 2: category URLs always include a fixed segment that cannot be edited, which would change every category address the brand has, and noindex can be set only for the whole site, not per template, so its filtered views and tag pages could not be kept out of the index. Platform B scores higher on editorial features and the product team prefers its editor, but the decision record lists the 2 failed must-haves with their consequences, and the brand chooses Platform A, budgeting for the add-on and for the redirect work the move will need.
Common mistakes
- Choosing on a demonstration theme. Speed and markup on an empty theme say nothing about a real catalogue.
- Trusting the feature list. "SEO-friendly URLs" can mean anything; test the capability.
- Ignoring generated pages. Tag archives, attachment pages and filtered views appear after launch.
- Assuming plugins will fill every gap. Some limits are in the platform's core and no add-on can change them.
- Leaving SEO out until migration planning. By then the platform's limits are the project's constraints.
- Forgetting redirects for the move itself. A platform that cannot hold the redirect map cannot be moved onto safely.
- No decision record. Without written results, the next team cannot tell which limits were known and accepted.
