What is technical SEO?
Technical SEO is the part of search optimisation that happens in the build rather than in the words. This guide explains what it covers, how it works according to Google's documentation, and what an SEO audit can and cannot tell you.
What does technical SEO include?
Google describes its search engine as working in three stages, and not every page makes it through each one. In crawling, automated programs download the pages they have found. In indexing, Google analyses each page and stores it in its index. In serving, it returns the indexed pages that are relevant to a search. Technical SEO is the work that carries a website's pages through the first two stages intact and removes the technical obstacles at the third.
In practice that means crawlers reach every page that matters, the right pages are indexed, each page has one address, the site describes itself in structured data, it is fast on a phone, it runs on HTTPS and language versions point to each other. Visitors never notice any of it when it works; when it fails, the first sign is often pages missing from search.
Technical SEO vs on-page SEO
On-page SEO is what each page says: the title and description shown in results, the headings, the text and the internal links. Content work decides which pages should exist, and off-page work concerns links and mentions from elsewhere. Technical SEO sits underneath all three: an excellent page with a forgotten noindex rule will not be indexed, while a technically flawless site with thin pages gives Google nothing worth showing. So the technical side is a foundation laid once and checked continually, and content is built on it month by month under ongoing SEO.
Crawling and indexing: how a page gets into Google
Google finds addresses by following links from pages it already knows and by reading sitemaps, then fetches each page and renders it with a recent version of Chrome, running its JavaScript. Links therefore need to be real HTML links: Google's documentation says it can only discover links that are anchor elements with an href attribute. Google also notes that server-side rendering or pre-rendering is still a great idea, because it makes a site faster for users and crawlers and not all bots can run JavaScript.
A page is eligible for indexing when Googlebot is not blocked, the page returns an HTTP 200 (success) status code and it has indexable content. Eligibility is not a promise: Google states that it does not guarantee to crawl, index or serve any page. In Google Search Console, the page indexing report shows which addresses are indexed and why others are not, and URL Inspection shows how Google sees a single page. When pages are missing, our guide on why a website is not showing on Google goes through the usual causes.
robots.txt, noindex and the sitemap
robots.txt is a text file at the root of the domain that tells crawlers which addresses they may request. Google says it is used mainly to avoid overloading a site with requests and is not a way to keep a page out of Google: a blocked address can still be indexed if other pages link to it. To keep a page out, use a noindex rule or a password, and do not also block that page in robots.txt, or the crawler never sees the noindex. Blocking rules carried over from a staging copy are a common launch mistake, covered in our technical SEO checklist for a new website.
A sitemap lists a site's pages, with details such as when each was last updated and which language versions belong together. It helps discovery but guarantees neither crawling nor indexing. Google says a site of about 500 pages or fewer, comprehensively linked internally, may not need one, though in most cases a site benefits from having one. Google ignores the priority and changefreq values that some tools still write, and uses lastmod only when it is consistently and verifiably accurate.
One address per page: canonicals and redirects
The same page can often be reached at several addresses: with and without www, over HTTP and HTTPS, with tracking parameters or through a shop's filters. Google says this duplicate content does not violate its spam policies, but crawlers may waste resources on addresses you do not care about, and Google is left to pick one version, the canonical, to show in results.
A site can state its preference, and Google ranks the methods by strength: a redirect and a rel="canonical" annotation are strong signals, inclusion in the sitemap a weak one. Google calls a canonical preference a hint, not a rule, and the methods work better combined, so they should agree: the canonical address should be the one in the sitemap and the one that redirects lead to.
When a page moves for good, Google recommends a permanent server-side redirect, HTTP 301 or 308; temporary redirects, 302 or 307, keep the old address in results. For a whole-site move it advises mapping every old address to a new one and keeping the redirects for as long as possible. A redesign is where this is most easily lost; our guide to redesigning a website without losing SEO covers the process.
What is schema markup?
Schema markup, or structured data, is a description of the page in code, using the schema.org vocabulary. Where the page tells a visitor that a business is an insurance agency with opening hours and contact details, the markup states the same facts in a form machines read without guessing. Google uses it to understand pages and, for the types it supports, to show rich results such as product details or breadcrumbs. Where the site's setup allows, it recommends the JSON-LD format.
Google's structured data policies add three points. The markup must describe content that visitors can see; marking up hidden or misleading content can lead to a manual action that removes the page's eligibility for rich results. Valid markup does not guarantee a rich result. And it should be tested: Google's Rich Results Test checks eligibility, and the Schema Markup Validator run by schema.org checks the vocabulary. On leadinsure.gr, which we designed, built and set up for search, the home page describes the business with the schema.org type InsuranceAgency, including its contact details and opening hours.
Core Web Vitals, mobile and HTTPS
Core Web Vitals are three measurements taken from real visits. Largest Contentful Paint (LCP) measures loading, with Google's target within 2.5 seconds; Interaction to Next Paint (INP) measures responsiveness, with a target under 200 milliseconds; Cumulative Layout Shift (CLS) measures how much the layout jumps, with a target under 0.1. Google's guidance is to measure at the 75th percentile: a page passes when at least three in four page loads meet all three targets, assessed separately on mobile and desktop.
Google says Core Web Vitals are used by its ranking systems, and also that it always seeks to show the most relevant content, even when the page experience is poor; chasing a perfect score only for SEO, it adds, may not be the best use of time. Speed is worth having for visitors first.
Google indexes and ranks the mobile version of a site, so text, structured data and metadata should be the same on mobile as on desktop; it recommends responsive design as the easiest pattern to implement and maintain. Whether pages are served securely, over HTTPS, is one of the questions in its page experience self-assessment. Keeping speed, certificates and updates in order after launch is the routine part of website care.
Language versions and hreflang
A site in more than one language should tell search engines which pages translate each other, through hreflang annotations in the page head, in HTTP headers or in the sitemap. Each version must list itself and every other version; if two pages do not point to each other, Google ignores the annotations. An x-default value names the page for languages that are not covered. On leadinsure.gr, wherever a Greek guide has an English counterpart, the two point to each other, with the Greek version as x-default; the Leadinsure case study covers the rest of the build.
Does technical SEO matter for AI search?
Yes, because it is the entry condition. Google says there are no additional requirements to appear in AI Overviews or AI Mode: to be shown as a supporting link, a page must be indexed and eligible to appear in Google Search with a snippet. The practices it lists are the familiar ones: crawling allowed in robots.txt and by any CDN or hosting layer, internal links, important content available as text, and structured data that matches the visible text. No new machine-readable files, AI text files or special schema markup are needed.
A separate robots.txt token, Google-Extended, lets a publisher decide whether content Google crawls may be used to train and ground its Gemini models; Google says it does not affect inclusion or ranking in Search. Other AI assistants run their own crawlers, and the same logic holds for them: a page a crawler may not fetch, or whose text appears only through scripts it does not run, gives it nothing to read.
What does a technical SEO audit include, and what can it not tell you?
A technical SEO audit examines a site the way a crawler does and compares the result with what the search engines themselves report. It checks status codes and broken links, redirects, robots.txt and noindex rules, the sitemap against the pages indexed, canonicals and duplicates, structured data, Core Web Vitals, mobile rendering, HTTPS, hreflang and the Search Console reports, including manual actions. The output is a prioritised list: what is wrong, why it matters and how much work it takes.
An audit also has limits. Third-party tools have no access to Google's internal ranking data, as Google itself points out, so a tool's score is not a ranking. An audit cannot predict positions or dates. A site with few visits may have no Core Web Vitals field data in Search Console at all, which leaves lab tests as an approximation. And it says little about whether the pages answer what customers search for; that is content work.
Google's advice to anyone hiring SEO help is that an audit should give realistic estimates of the improvement and of the work involved, that the auditor needs only read access to Search Console at that stage, and that you should look elsewhere if anyone guarantees first place. The audit is the first step of our technical SEO setup, and it is where we start before taking on the maintenance of a website we did not build.
Frequently asked.
Is technical SEO a one-off job?
The foundation is set once, ideally before launch, but it does not stay set. New pages, plugin or template updates, a change of hosting or a new language can each undo part of it, so technical monitoring continues under a maintenance plan or an ongoing SEO programme.
When does a website need technical SEO work?
At launch; before a redesign or a move to a new platform or domain; when a language is added; and when the page indexing report in Search Console shows important pages as not indexed. Google itself names a redesign or a new launch as a good time to bring in SEO help, the earlier the better. A site left without maintenance for a long time is also worth an audit, because problems accumulate quietly.
Can I check the basics myself?
Yes. Verify the site in Google Search Console and read the page indexing and Core Web Vitals reports, run URL Inspection on your key pages, open your domain followed by /robots.txt to see what it blocks, and look at every key page on a phone. These checks show symptoms; the causes usually need a look at the code.
Do I need an llms.txt file for AI search?
Not for Google. Its documentation on AI Overviews and AI Mode says no new machine-readable files, AI text files or special schema markup are needed. The same technical foundation serves classic results and AI answers.
How much does technical SEO cost?
It depends on the number of page templates, the platform, the state of the code, any redirects from an old site and the number of languages. A recently built five-page site with one form needs a short check; a long-running shop with thousands of product addresses, filters and two languages needs a page-by-page plan. We quote every project in writing after a conversation, and our guide to what SEO costs and how long it takes explains the factors.
Can technical SEO guarantee a first-page ranking?
No. Google does not guarantee to crawl, index or serve any page, and it says that no one can guarantee a first-place ranking. Technical work removes the obstacles; relevance, content and competition decide the positions.