Search Engine Optimization (SEO)

Search Engine Optimization is the discipline of aligning a page with the way a search engine crawls, indexes and ranks. A search engine does three jobs: it discovers URLs by following links and sitemaps, it stores and parses what it finds, and it orders the stored pages against each query a user types. SEO is the work of removing friction from each of those jobs — so the page is found quickly, parsed correctly, and judged the best available answer.

The practical definition we work to: a page is search-optimised when a crawler can fetch it without obstacles, a parser can tell exactly which question it answers, and a reasonable person comparing it to the other results would agree it deserves to be among them. Note what is absent from that definition: keyword density formulas, exact-match domains, and word-count targets. Those are artefacts of how ranking worked a decade ago, not how it works now.

Modern ranking systems model the query and the document in the same semantic space, which means a page can rank for a phrase it never uses, provided it covers the underlying topic well. The implication for writers is liberating and demanding at once: you no longer need to repeat the phrase, but you do need to genuinely cover the subject.

Strip away the mythology and the pipeline has four stages. Crawling: Googlebot fetches URLs it has discovered, at a rate shaped by your server's responsiveness and your site's perceived value. Indexing: the fetched page is parsed, rendered if necessary, and stored — or dropped if it is a duplicate, blocked, or judged worthless. Retrieval: for a given query, the index returns every document that plausibly matches. Ranking: those candidates are ordered by hundreds of signals, of which the durable ones are relevance of the content, the authority implied by links from other trusted sites, page experience, and the searcher's context such as location and language.

Two consequences follow. First, each stage is a filter, so failure is silent: a page that is never crawled, or is crawled but not indexed, produces no error message anywhere except a quiet absence from results. Search Console exists precisely to make those silent failures visible, and it is the first tool to connect to any site you care about.

Second, the stages interact. Strong authority cannot rescue a page that is not indexed, and perfect technical hygiene cannot lift a page nobody would choose to read. This is why productive SEO work always starts with a diagnosis of which stage is failing, rather than with a generic checklist applied everywhere at once.

The first pillar is technical access. The page must be reachable by a crawler, permitted by robots.txt, free of noindex directives, served over HTTPS, fast enough not to time out, and identified by a single canonical URL. None of this earns rankings — it merely qualifies the page to compete. Sites fail here far more often than their owners believe, usually through a staging rule or a template change nobody audited.

The second pillar is relevance. The content must actually resolve the query a searcher typed, at the depth the query deserves. A definitional question wants a direct definition in the first paragraph. A comparison query wants a real comparison with criteria and a verdict. A transactional query wants the product and its price. Matching the format to the intent matters as much as matching the topic.

The third pillar is authority. Links from other credible sites remain the strongest evidence that a page is worth trusting, because they are the signal hardest to fake at scale. Authority accrues to pages people cite — original data, useful tools, clear explanations — and it flows through internal links, so a site with a strong homepage and orphaned content is leaving most of its authority unused.

SEO optimises for a position in a ranked list of links. Answer Engine Optimization optimises for a passage being extracted and quoted by an AI assistant. Generative Engine Optimization optimises for being the source credited when an engine composes an answer. The three share the same technical foundation — crawlable, server-rendered, canonical content — but diverge in what they reward at the margin.

The practical relationship is sequential. A page that is invisible to classical search is almost never cited by an AI engine either, because the retrieval pools overlap heavily. Fix retrieval and relevance first; the extraction and attribution work sits on top of it. Our guide to how the three disciplines differ goes deeper, and the readiness audits on this site check both layers separately because sites usually fail at only one.

Start with Search Console, not with content. Verify the site, submit the sitemap, and read the coverage report. The single most common finding on a new site is that a meaningful fraction of its pages are excluded for a reason the owner never saw — a stray noindex, a redirect chain, a duplicated canonical. Fixing retrieval beats writing anything new.

Then pick a small set of queries you genuinely deserve to win — specific, lower-competition phrases where your page can be the best answer on the internet — and make each one the job of exactly one page. Give each page a descriptive title, a single clear H1, a meta description worth clicking, and content thorough enough that a reader needs no follow-up search.

Finally, build the habit of earning links rather than asking for them: publish the calculation, the dataset, the tool or the explanation that others in your field will reference. That is the slowest part of SEO and the one that compounds longest.