How to find and fix technical SEO issues on your website

Technical SEO goes wrong in a specific order of severity, and almost nobody works through it in that order. People start with whatever a tool made easy to see, such as image alt text and heading levels, while a canonical tag quietly points the whole site at its homepage.

What follows is a workflow sorted by consequence. Work top to bottom and stop when you run out of time, because everything below the line you stopped at genuinely matters less than everything above it.

The four tiers, and why the order is fixed

The reason the order is fixed rather than a matter of taste: a Critical fault makes every Medium improvement to that page worthless. Optimising a heading structure on a noindexed page is work with a guaranteed return of zero.

Tier 1, Critical: is the page reachable and keepable?

Six checks. On most sites all six pass, and on the sites where one fails it is usually the only thing that matters.

robots.txt is blocking something it should not

Fetch /robots.txt and read it. The specific thing to look for is a Disallow: / left behind when a staging configuration reached production. The subtler version is a directory rule that catches more than intended, because robots patterns are prefixes rather than paths — Disallow: /blog also blocks /blogging and /blog-archive.

Two behaviours catch people out. The most specific rule wins regardless of file order, so moving lines around changes nothing. And a crawler obeys exactly one group: once a Googlebot group exists, the * group is ignored entirely, even if the Googlebot group is about something unrelated.

A noindex directive is still in place

Check two places, not one. The meta robots tag in the page source, and the X-Robots-Tag HTTP header, which does the same job invisibly and is the version people miss for months. In WordPress there is also a single checkbox under Settings → Reading that applies noindex site-wide.

The canonical points at the wrong page

A canonical tag tells Google which URL is the real one. A template emitting the homepage URL on every page asks Google to index the homepage instead of any of them, which removes a whole site from search while every page still loads perfectly for visitors.

The server returns the wrong status code

A missing page that returns 200 with an error message on it is a soft 404, and Google has to guess. A page returning 5xx intermittently gets dropped. And robots.txt itself returning a 5xx is the one genuine emergency in this area: Google's documented behaviour is to treat a persistently failing robots.txt as though the entire site were disallowed.

Redirects loop or chain

A loop makes a URL unreachable outright. A chain of three or four hops wastes crawl budget and loses a little at every step. Both are invisible in a browser, which follows them silently and shows you the destination.

The content only exists after JavaScript runs

Google renders JavaScript, but rendering is a second pass that can fail or be delayed. Press Ctrl+U to view source and compare it with what you see. If the source is an empty div and the words you want to rank for arrive later, you are betting on that pass. Get the title, headings, body text and links into the server's response.

Tier 2, High: what splits or wastes your indexing

These do not stop a page existing. They stop it counting for as much as it should.

• Duplicate or near-duplicate pages. Two pages covering one topic do not get two chances. Google picks one and discounts the other. Filtered ecommerce URLs and city pages that differ only by city name are the usual sources.
• Parameter URLs multiplying. Sorts and filters generating thousands of crawlable addresses spends your crawl budget on combinations nobody searches for.
• Orphan pages. A page linked from nowhere on your own site is barely a page. A sitemap entry is a suggestion; an internal link is a vote.
• Broken internal links. Each one wastes a crawl and signals neglect. External ones are less costly and still worth clearing.
• The sitemap disagreeing with the site. Listing redirects, 404s or noindexed URLs makes the whole file less trustworthy.
• Mixed http and https, or www and non-www, both resolving. That is duplication at the domain level, which is the most expensive kind.

The number worth writing down at this tier: how many URLs Search Console reports as indexed, against how many pages you believe you have. If you have 80 pages and Google reports 4,000, something is generating URLs. If it reports 12, something is preventing discovery. Both are invisible from the front of the site.

Tier 3, Medium: how well an indexed page performs

Real improvements with smaller, slower returns. Worth doing once the tiers above are clear.

• Unique, descriptive titles on every page, with the meaningful words first. Duplicated titles across a site are a common template fault.
• Meta descriptions that are distinct and worth clicking. Not a ranking factor, but they are the advertisement for the click.
• One H1 per page, and a heading structure that does not skip levels for styling reasons.
• Core Web Vitals in the passing range, measured from field data rather than a lab score.
• Images sized to their display size and served in a modern format, with width and height attributes set.
• Structured data that matches the visible content. Marking up something not on the page is a policy violation rather than a shortcut.

On performance specifically, measure the right thing: Core Web Vitals are assessed from real visits at the 75th percentile over a rolling 28 days, so a PageSpeed score of 68 is a diagnostic and not a finding. Why your website is slow covers which of the four underlying causes you actually have.

Tier 4, Low: things that are not problems at all

Worth knowing so you can stop paying attention to them.

Running the whole thing in about an hour

1. 1Open Search Console → Pages. Compare indexed against not-indexed and read the reasons. This single report finds more than any crawler.
2. 2Run URL Inspection on your three most commercially important pages. Fix anything reported as blocked, noindexed or canonicalised elsewhere.
3. 3Fetch robots.txt and read it in full. It is thirty seconds and it is where the worst faults hide.
4. 4Audit those same three pages for the Critical and High items above.
5. 5Crawl the whole site once to find duplicate titles and broken links in bulk.
6. 6Only then start on Tier 3.

When to run this

A full pass twice a year is enough for most sites. What is worth doing monthly is much smaller: check Search Console for new indexing errors and confirm nothing important started returning a 404.

Run the Critical tier immediately after any redesign, platform migration or domain change. Those three cause more sudden traffic losses than algorithm updates do, and the damage is nearly always an unfinished redirect map or a staging noindex that shipped. The website redesign checklist covers what to verify on launch day.

Questions people actually ask

How do I know which tier a problem is in?

Ask one question: does this stop the page being crawled, indexed or served? If yes, it is Critical. If it splits or wastes indexing, it is High. If it only affects how well an already-indexed page competes, it is Medium at most.

My audit tool found 200 issues. Where do I start?

With the Critical tier, which in practice is fewer than ten items. Clearing those usually moves more than the other 190 combined, and most of the 190 are Tier 4 findings that read as problems and are not.

Do I need a developer for this?

For diagnosis, no: every check above is something you can run yourself. For fixes, it depends: robots.txt and canonical tags are usually settings, while status codes, headers and rendering are code. If the site fights you at every change, that is its own finding — ten signs a website needs rebuilding covers when the platform is the constraint, and our website maintenance work exists for the case where it is not.