SEO, GEO and Google Ads for an online derealization and depersonalization therapist

In short

Ben Meijer is a therapist who treats derealization and depersonalization (DPDR) online. He wanted more bookings through Google and to be mentioned more often by AI systems such as ChatGPT and Google AI Overviews. I fixed the technical foundation of his Dutch and English sites, wrote content around the questions his audience actually asks, and repaired his Google Ads conversion tracking. By September 2026 both sites ranked higher on average, and in the US Google named him first for “Best DPDR therapist online”.

Facts at a glance

About the client

Ben has been a therapist since 2003 and now works only with DPDR. He works online over Zoom, with clients in the Netherlands and abroad. Many of his visitors search at night in a panic, for example after a bad trip. That makes this YMYL content (Your Money or Your Life: topics where Google is extra strict because they affect someone’s health).

The question

Ben wanted to reach more clients. He wanted more bookings through Google and to know whether AI systems knew him. He was also running an Ads campaign without seeing what it delivered.

The starting point

I began by crawling both sites and going through the Search Console data. On the English site, an old staging copy was open to Google and showed up for dozens of keywords instead of the real site. On the Dutch site, eight pages competed for “what is derealization”, so none of them reached the top 10. Structured data (code that tells search engines who you are and what you do) was missing or outdated. In a baseline test in June, Google AI Overview did not use the Dutch site as a source once across 4 test questions.

What I did

Technical foundation for Google and AI (June and July 2026)

I set the staging copy to noindex and had it removed from Google. I added schema for the person, organization, FAQ and articles, linked to Wikidata and both YouTube channels. I also added llms.txt (an information file for AI systems), allowed AI crawlers in robots.txt, and set up hreflang (the language link between the two sites).

Structure and content

Instead of eight partial answers, I chose one main page with contextual links from the others. I rebuilt the anhedonia page (anhedonia: no longer being able to feel anything) with a direct answer at the top and an FAQ. The August AI test showed that Ben was never named for questions like “will derealization make me go crazy”. I wrote new content for those questions, which Ben edited for accuracy. I deliberately did not duplicate that text on the FAQ page: one full answer in one place, short references elsewhere.

Google Ads

My first suspect was the cookie banner, but checking ruled it out. The real cause was a form trigger waiting for a thank-you message that never appeared on the site. That conversion had not received a single signal since March 2023. WhatsApp clicks and bookings through the web shop were not tracked either. I fixed the tracking first and then adjusted the campaign. I only move on to structure and copy now that the tracking is right.

How I checked it

• After every change in WordPress I reload the page and check the revision. Then I clear the cache and open the live page with a cache buster (an extra bit in the URL so you never see an old copy).
• I validated every schema before publishing and tested it again live in the Rich Results Test. Before saving, I check that the JSON contains no backslashes.
• I checked the staging copy for noindex on several pages. In September a site: search returned zero results.
• I tested the Ads tracking with a fake click ID in the URL. A phone click, a WhatsApp click and a test form all sent a conversion signal with status 200 and the click ID attached. I verify every account change afterwards in the change history.
• I measure AI visibility monthly with fixed test questions per language on four platforms, through a VPN in the Netherlands and the US.
• When something looks off, I trace the source. One AI system gave the wrong town as his practice location. That came from the registered address in the footer and from two old client stories.

Results

Google Search Console, 1 to 16 September 2026 compared with 1 to 16 July 2026:

• Dutch site: average position from 11.9 to 10.4. “derealisatie” from 15.3 to 11.8, “ben meijer” from 7.8 to 4.5. Clicks did go down in that period.
• English site: average position from 17.6 to 12.4, with the same number of clicks.
• Impressions in AI Overviews and AI Mode (Search Console report on Google’s AI features): Dutch roughly unchanged, English up by about a third. The Dutch anhedonia page was shown more than four times as often.

My own AI test, 8 and 10 September 2026: the Dutch site was used as a source in 24 of 80 checks. In June it was 0 of 4. These tests are not directly comparable, because September covered more questions and platforms. In the US, Google AI Overview named Ben first for “Best DPDR therapist online”. For the English site, ChatGPT and Perplexity do not mention him yet.

Google Ads now tracks forms, phone clicks, WhatsApp clicks and web shop bookings.

What’s next

The work continues. Once Ben approves, I take the next steps in Google Ads and in the content, and I keep measuring every month.

Frequently asked questions

How quickly do AI systems start naming a practice?

Here the technical foundation was in place in July, and I saw the first mentions in September. A new page appeared in AI answers within a few weeks, but without Ben’s name yet. Being shown and being named are two separate steps.

Is Google Ads worth it if conversion tracking is broken?

Not much. The campaign is flying blind. Here Google only counted phone clicks, while bookings came in through the web shop. Tracking first, then the rest.

Does an English site need its own approach?

Yes. Google picked up the English site, ChatGPT and Perplexity have not yet. You only see that with separate test questions per language.