SEO · 03
Technical SEO Consulting.
Google often sees less of your site than your users do. I find what it misses, specify the fix per page type, and tell you when the cause is not technical.
You hire me when
- Google indexes the wrong pages, or less than you publish.
- Your site runs on React, Vue or Angular.
- Traffic dropped and nobody can say why.
- You want specs and tickets, not a PDF.
What you get
- The three SEO tests on every page type, re-run as proof.
- Markup specs: the HTML each page type must ship.
- JavaScript and release rules Google can render.
- Page-weight and compression budgets with a check per deploy.
- Structured data fixed and tested, including job and author markup.
- Audits as prioritised tickets.
Why Google sees less than your users
Google does not look at your site the way a browser does. It fetches the raw HTML first. It renders the page later, with JavaScript and CSS files it may have cached long ago. It follows only real links.
Google on its renderer: “WRS may ignore caching headers. This may lead WRS to use outdated JavaScript or CSS resources.” (Google Search Central, JavaScript SEO basics)
So a page can look perfect in every browser and still show Google an empty list. As I put it in a 2026 workshop: in the HTML, everything that is not structural markup and content is ballast. My goal: Google gets the same content as your users, in the first HTML, small and linked, on every release.
What goes wrong in 2026
Missing titles and broken canonicals still happen. They are rarely where the traffic goes. These are the failures I measured in my 2026 audits, and the fix I specified for each:
| What breaks | What I measured | What I specify |
|---|---|---|
| JavaScript removes what the server sent | Google’s own test: header, H1, footer, 0 listings | Hashed file names, old files kept online, client code never removes server content |
| Pages too heavy for what they show | 1 percent text in a 900 KB list page; almost every page type over 250 KB | 150 KB of unzipped HTML, 250 KB at most |
| Compression that makes pages bigger | Brotli bigger than gzip, page type after page type | One CDN setting, one check per deploy |
| Content only after a tap | 0 FAQ answers in the HTML | Unique content in the HTML, visible |
| Markup Google cannot use | every audited job ad without a location Google can read | The missing fields, from data the site already has |
| Crawl paths that go nowhere | 0 paginated list pages seen by Google in 16 months | Real links with href, one URL per page |
JavaScript removes what the server already sent
The server sent each list page with its listings. Then the JavaScript loaded a file from another release, crashed, and took the listings off the page.
- File names stayed the same across releases, so a page Google had stored with one release got the JavaScript of another.
- In Google’s stored copy of 4 in 10 top list pages, the JavaScript had crashed.
- In October I ran Google’s own Rich Results Test on a top list page. Google’s screenshot: header, H1, footer. 0 listings. In my replays, one missing file was enough.
Google names the fix itself: “Content fingerprinting avoids this problem by making a fingerprint of the content part of the filename” (source).
The defect is proven in the code and in Google’s renderer. How many clicks it cost is not measured, so I do not print a number for it.
What I specify:
- A content hash in every file name. Old files stay online for months after a deploy.
- An error boundary around every part that loads late.
- Client code never removes server-rendered content.
Pages too heavy for what they show
A list page shipped about 900 KB of HTML. 1 percent of it was text. The rest was framework data, inline CSS and inline icons. Almost every page type was over 250 KB.
I trimmed a real page without changing what it shows: from over 900 KB to about 200 KB, with no visible change.
List pages carried over 2 MB of HTML, most of it a data blob with far more listings than the page shows. A lean rebuild of the same page went from 1.5 MB to about 120 KB, and its Lighthouse score from under 30 to over 90. The weight alone was not the cause there. It is risk, and cheap to remove.
What I specify: 150 KB of unzipped HTML per page, 250 KB at most. Compression hides the weight from the network, not from the device.
Compression that makes pages bigger
Brotli should be smaller than gzip. Here it was bigger, page type after page type. Googlebot asks for Brotli, so it got the heavier file.
The quality level was not the cause. I recompressed the same page at every level, and every one beat the live file. The cause was a streaming setting at the CDN that compressed the page in tiny pieces. Set right, the page is much smaller than what was served.
What I specify: one CDN setting, no template change, and a check on every deploy: Brotli bytes smaller than gzip bytes.
Content that appears only after a tap
Product pages had a row of FAQ buttons and no answers in the HTML. About a third of the page’s text sat behind those taps. On the FAQ pages, every answer existed only inside a script.
What I specify: unique content in the HTML, visible without a tap. Text that repeats on every page may load on tap.
Markup Google cannot use
Job ads with structured data, and hardly any clicks from Google’s job search.
- In every audited ad, Google could not read the location. For Google, city, region and country were empty.
- Salary and employment type: in the markup of 0 ads.
- A job board republished the same ads with four more fields. Google showed the job board as the source.
Google on its job search: “The new user experience enables job seekers to filter by various criteria like location or job title” (Google Search Central, Job posting structured data). An ad without a location Google can read cannot match a location filter.
On one site, the logo in the markup was blocked for Google’s image crawler.
What I specify: the missing fields, filled from data the site already has, each tested in the Rich Results Test.
Crawl paths that go nowhere
- Google found no crawlable link to page 2 of any list. In 16 months, 0 paginated URLs got a single impression, and Google called page 2 “URL is unknown to Google”.
- On another site, the link to older entries was no real link. Google could not follow it.
- In a sample of Googlebot’s HTML fetches, only about 40 percent hit a canonical, indexable page. Dozens of ad-parameter copies of the home page were crawled too.
Google: “Generally, Google can only crawl your link if it’s an <a> HTML element (also known as anchor element) with an href attribute.” (Google Search Central, Link best practices)
What I specify: a real <a href> for pagination and every path you want crawled, and one URL per page. A canonical is a safety net, not a fix.
How I work
- The three SEO tests on every page type. PageSpeed Insights, the page with JavaScript off, and Google’s live test. In one 2026 audit: about 50 URLs, 30 tools, 600 findings.
- Every tool finding reproduced by hand. A crawler is a hypothesis generator, never a verdict. On one home page, half the warnings from a good tool were false: it read a HEAD request and missed the Brotli of the real page.
- Field data first, lab runs repeated. One PageSpeed run is not a number. 10 runs of the same page scored from under 40 to over 90.
- Google’s own tools on your live pages. In one case: scheduled Rich Results Tests, and a watcher that fetched about 150 list URLs every 30 minutes, to catch the failure while it happened.
- Ruling out, too. One site lost a large share of its Google traffic. The page with JavaScript off passed on every page type, and every page type was indexed. The cause was not in the code, and I said so.
- Specs, not a PDF. Every finding ends as the HTML a page type must ship, a budget, and a check your team runs on every release. As I wrote on a 2026 slide: “A process is worth a 1000 tasks.”
This site, measured the same way
This site is static HTML on Cloudflare Pages since 21 October 2026. On 24 October 2026 I ran my own checks on all 100-plus URLs in its sitemap.
- Every one answers 200. Every page is static HTML: with JavaScript off, nothing is missing.
- Median about 40 KB of unzipped HTML. One page breaks my own 150 KB budget: the workshop list, at about 180 KB. None is over 250 KB.
- PageSpeed Insights, mobile, on three pages: performance 95 to 98, SEO 100, lab LCP 2.4 to 2.6 seconds.
- Brotli is bigger than gzip on a handful of pages, and only about 1 percent smaller over all of them. The same failure I find in audits, at small scale, on my own CDN.
Run PageSpeed Insights on this page and check my numbers.
Questions I get in every technical audit
Does Google render JavaScript?
Yes, later, with files it may have cached. Google: “Keep in mind that server-side or pre-rendering is still a great idea because it makes your website faster for users and crawlers, and not all bots can run JavaScript.” (source) So the main content belongs in the server HTML.
Is our page too big for Googlebot?
Google: “When crawling for Google Search, Googlebot crawls the first 2MB of a supported file type” and “The file size limit is applied on the uncompressed data.” (source) My budget is far lower, 150 KB, because the weight costs your users first.
Should deleted pages answer 404 or 410?
Either. Google: “All 4xx errors, except 429, are treated the same” (source). A 410 is not faster. What matters is that a page with no content does not answer 200.
Do we need a full audit first?
No. The three tests on one URL per page type find most problems in a day. The full audit comes when the tests show where to dig.
What do our developers get?
Specs per page type: the HTML each page must ship, a page-weight budget, and the checks to run on every release. Written as tickets.
Sources for the Google quotes on this page, read on 2026-10-24: JavaScript SEO basics, Googlebot, Link best practices, Job posting structured data, Google Search Central, and How HTTP status codes affect Google’s crawlers. The client numbers come from my own audits, workshops and handovers in 2026.
From real jobs
- 2026: JavaScript removed server-rendered listings, caught live in Google's own test.
- 2026: Brotli bigger than gzip, page type after page type; one CDN setting was the cause.
- 2026: about 50 URLs, 30 tools; one list page trimmed from over 900 KB to about 200 KB with no visible change.
- 2026: every audited job ad without a location Google can read.
- 60-plus sites through the SEO tests since 2016.
- A page-type matrix for automated landing pages.
- A lazy-loading spec that keeps every piece of content on one URL.
Teams like yours
Some of our customersOnline shops
Webshops and marketplaces: category pages, product pages, speed.
ExamplesMarketplaces and job platforms
Real estate, jobs, cars: list pages decide traffic.
ExamplesTravel and tourism
Marketplaces, operators, tourism boards: many markets, many languages.
ExamplesSoftware, SaaS and apps
SaaS and apps: JavaScript, docs, growth.
ExamplesSomething else in the way?
How consulting works- External SEO lead
- Relaunch and migration SEO
- International SEO
- Web performance
- SEO for newsrooms
- Content strategy and search demand
- Search data and traffic forensics
- SEO agency cleanup
- GEO and AI search
- Vibe coding training
- AI automation consulting
- AI-first transformation
- Custom tools for one job
- Design thinking
- Growth
- Growth strategy analysis
- SEO workshops and training
- Keynotes and talks
- Sparring and second opinion
- Dev, product and marketing mediator