Where SEO work actually happens on a website

SEO isn't a page. It's eleven places

"Where do I do SEO?" is a fair question that almost nobody answers literally. The advice is usually a list of activities — write content, build links, improve speed — which is useless if you're staring at a WordPress dashboard wondering which box to type in.

So here's the literal answer. These are the eleven physical locations where an SEO change lands, what lives in each, and whether you can touch it yourself.

The head tag: four lines that decide whether you show up

The head is invisible to visitors and the first thing a crawler reads. Four elements matter more than everything else in it combined.

You can see all four right now: open any page, press Ctrl+U, and read the first thirty lines of source.

• Title tag — the strongest on-page signal you control and the line people decide to click. Write it for the query, not for your brand. Google rewrites title links often, and it rewrites the bad ones most.
• Meta description — not a ranking factor, but it is your ad copy in the results. Google frequently replaces it with page text when yours doesn't match the query.
• Canonical tag — tells Google which URL is the real one when several show the same content. It's a strong hint, not an order. What a canonical tag does covers where it goes wrong.
• Meta robots — noindex keeps a page out of the index; nofollow stops link signals passing. A stray noindex left over from a staging site is one of the most common causes of a page vanishing.

The body: headings, copy, images and the links between pages

This is the part you can fix without asking anyone. It's also where most of the winnable ground is, because most sites treat the editor as a word processor.

Heading order is the cheapest fix on any site. One H1 per page, matching what the page is actually about. H2s that read as the sub-questions a searcher would ask — not as brand slogans. If your H2s are 'Our Approach' and 'Why Choose Us', you have decorated a page rather than answered anything.

Internal links are the second cheapest. Every new page needs links from pages that already have authority, and a page nobody links to internally is one Google struggles to justify crawling. Our internal linking best practices has the specifics.

Images need three things: a descriptive filename, alt text that describes the image rather than stuffing a keyword, and explicit width and height so the layout doesn't jump. The first two are yours. The third usually isn't.

Templates versus pages: where one edit hits four thousand URLs

This is the distinction people miss, and it's the one that decides whether SEO work scales or crawls.

A page edit changes one URL. A template edit changes every URL that inherits it. On an ecommerce site, fixing the product template's title pattern from Product Name | Store to something including the category or model number changes thousands of titles in one deploy. Writing them one at a time is a quarter of somebody's life.

It applies in reverse too. A broken template is a site-wide problem wearing a page-shaped disguise — a missing canonical in the product template creates duplicate URLs across the whole catalogue, and no amount of page-by-page editing will fix it.

• Do at template level: title and meta patterns, schema markup, breadcrumbs, canonical rules, pagination handling, image loading.
• Do at page level: the actual copy, the H1, in-body internal links, FAQ blocks, alt text, anything that should be genuinely different per page.
• Never do at template level: meta descriptions that end up identical across 400 pages. Google will ignore them and so will readers.

The files nobody opens: robots.txt, sitemap.xml and the server

Three locations sit outside your CMS and outside most people's awareness. All three are checkable in about five minutes.

robots.txt lives at yoursite.com/robots.txt. Open it. You're looking for a Disallow line that blocks something important — /blog/, /products/, or in the worst and surprisingly common case, Disallow: / left behind from launch.

The XML sitemap usually sits at /sitemap.xml or is linked from robots.txt. Open it and check it lists the URLs you want indexed and only those. Sitemaps that include redirected URLs, 404s or noindexed pages send Google a confused signal about what you consider canonical.

Server configuration is where redirects and status codes are decided, and status codes are not a CMS setting. A page that shows a 'not found' message to humans while returning 200 to crawlers is a soft 404, and Google treats it as a quality problem rather than a missing page. Check with a browser extension or a crawler — never by looking at the screen.

1. Open /robots.txt and read every Disallow line out loud.
2. Open your sitemap and spot-check ten URLs for redirects and 404s.
3. Run a crawl and sort by status code. Anything that isn't 200 or an intentional 301 needs a decision.
4. Check HTTPS and the www/non-www version both resolve to one canonical host with a single redirect hop.

The layer above your site: CDN, edge and rendering

This one barely existed as an SEO concern a decade ago and now routinely decides whether a site gets indexed at all.

A CDN can hold redirect rules, cache HTML, rewrite headers, resize images and block traffic it thinks is a bot. Every one of those can help or quietly break things. Undocumented edge redirect rules are a classic cause of chains that make no sense in the application code.

Rendering is the other half. Google executes JavaScript, but in a second pass and on its own schedule. Content that appears only after a click or an infinite scroll may never be seen. If your main content, internal links and title tag exist only after JavaScript runs, you're asking Google for a favour rather than giving it a page.

The test is free: use the URL Inspection tool in Search Console, request the live URL, and read the rendered HTML Google actually got. If your product description isn't in there, nothing else on this page matters.

Which changes need a developer, and how to ask

Split the work honestly and you'll ship more of it. Roughly half of the eleven places are yours; the rest need someone with deploy access, and vague tickets are why those changes sit in a backlog for six months.

A good ticket names the exact URL pattern, the current behaviour, the required behaviour and the test. "Improve SEO on the blog" gets ignored. "All /blog/ URLs currently return 200 with a canonical pointing at the homepage; they should self-canonicalise. Test: view-source on any three blog posts shows its own URL in the canonical" gets done on Thursday.

One more thing worth saying: not all SEO happens on your website. Your Google Business Profile, the sites that link to you, and the marketplace or directory listings that outrank your own pages are all places where the work lands. You just have less control over them, which is exactly why the eleven places you do control deserve to be right first. If the technical half of that list is beyond your team, it's what technical SEO work covers.

• You, today: titles, meta descriptions, H1 and H2 order, body copy, alt text, in-body internal links, FAQ blocks.
• You, carefully: URL slugs (you owe a redirect), robots.txt via a plugin, sitemap settings.
• Developer required: templates, schema markup, redirects and status codes, response headers, image formats and loading, CDN and edge rules, rendering strategy.