Technical SEO for company websites

What technical SEO does and what it doesn't

Technical SEO answers three questions: Can a search engine reach the page? Can it read the content? Can it understand it?

Nothing more. It doesn't make bad copy good or a missing page visible. It makes sure existing content gets its chance.

That is why the order matters: first make sure there is something to find, then make sure it gets found. A technically flawless website with three pages ranks for three topics.

The reverse is also true: a technical error can make good content completely invisible. An accidentally set noindex directive removes a page from the index entirely, no matter how good it is. That makes technical SEO a prerequisite, not a lever.

The ten-second test

Technical SEO answers three questions: Can a search engine reach the page? Can it read the content? Can it understand it?

Nothing more. It doesn't make bad copy good or a missing page visible. It makes sure existing content gets its chance.

That is why the order matters: first make sure there is something to find, then make sure it gets found. A technically flawless website with three pages ranks for three topics.

The reverse is also true: a technical error can make good content completely invisible. An accidentally set noindex directive removes a page from the index entirely, no matter how good it is. That makes technical SEO a prerequisite, not a lever.

The six points

1. Indexability. Every page that should be found must not carry noindex and must not be blocked in robots.txt. Check: Search Console, page indexing report.

2. Sitemap. An XML sitemap that contains all relevant pages and no redirects or blocked URLs. Submitted in Search Console.

3. Title and description. Unique per page. Title under 60 characters with the main term at the front, description 120 to 160 characters.

4. Heading structure. Exactly one H1 per page, below it H2 for the sections. No headings used as design elements.

5. Load time. Above all, Largest Contentful Paint under 2.5 seconds on a smartphone. The most common drag is images that are too large.

6. Structured data. JSON-LD for Organization, Article and BreadcrumbList. This tells machines what a piece of content is.

If you have these six right, you are technically further ahead than the vast majority of competitors.

The mistake no tool reports

The most consequential technical mistake doesn't show up as an error in any standard report: content that is only created in the browser by JavaScript.

In the browser, everything looks right. In the delivered HTML, there is nothing in that spot. Search engines can partly compensate because they render pages, but with a delay and not always completely. The fetch tools of generative systems usually don't do it at all.

Typically affected are expandable sections, tabs, lazy-loaded lists and anything that comes from a CMS via an API.

The test takes ten seconds: open the page, view the source code, use the search function to look for a sentence from the middle of the text. If it isn't found, the content isn't in the document.

If you want to check an entire website, fetch it without JavaScript and compare the amount of text. The difference is usually clear.

Load time: what really matters

Of the Core Web Vitals, one is decisive in everyday practice: Largest Contentful Paint, the time until the largest visible element has loaded. Target value under 2.5 seconds on a smartphone.

In more than half of cases, the cause is the images. Three measures have an immediate effect:

Format. WebP instead of JPEG or PNG typically saves 30 to 50 percent at the same quality.

Size. An image displayed at 600 pixels wide doesn't need to be delivered at 3,000 pixels wide. This is the most common and most expensive mistake.

Loading behavior. Only load everything below the visible area when it is needed. The main image at the top, on the other hand, should be loaded with priority, not delayed.

After that come fonts (limit the number of weights, serve them locally) and external scripts. Every tool loaded in the head delays everything after it.

Cumulative Layout Shift is the second value that counts and is easy to fix: specify width and height on every image so nothing jumps.

What a health score of 100 doesn't mean

Many audit tools show an overall score. A score of 100 with zero errors feels finished, and it regularly hides the more important findings.

The reason: these tools rank by technical severity, not by business impact. A missing H1 is a warning, not an error. So is a description that is too short. Both can still mean that Google builds the snippet itself and the click-through rate drops.

That is why it pays to look beneath the overall score:

How many indexable pages have no H1?

How many have a description that is too short or missing?

How many titles are longer than 60 characters?

How many pages have no internal links?

These are the numbers that say something about the state of the site. The overall score only says that nothing is broken.

The order of implementation

Week 1: Ensure visibility. Set up Search Console and check which pages are indexed. Submit the sitemap. Fix incorrectly set noindex directives and robots.txt blocks. Without this step, nothing after it works.

Week 2: Readability. Do the HTML test for every page type. Where content is missing, clarify with development. This is the only point that can get technically complex.

Week 3: Meta data and structure. Title and description per page, check the H1. This is diligence, not expertise, and it has a direct effect on the click-through rate.

Week 4: Load time. Convert and shrink images, set dimensions, review scripts.

Week 5: Structured data. Organization and WebSite centrally, Article and BreadcrumbList via the template.

After that, a check once a quarter is enough. Technical SEO is not a permanent construction site but a state you create once and then maintain.