ChatGPT Already Sends Us More Conversions Than Google SEO
A measured look at 28 days of GA4 for a tutor CRM: ChatGPT drove more conversions than Organic Search, while domain migration still holds Google equity on the old host. What we are doing next.
In short: Over 28 days (Jul 31–Aug 28 2026), ChatGPT already sent OBRI more GA4 conversions than Organic Search (5 vs 2), with most AI sessions landing on /en and /. Google equity is still settling after our move from studentportalua.com to obricrm.com. So we are doubling down on extractable product pages and LLM-first writing — not on more thin programmatic URLs.
I build OBRI, a CRM for independent tutors. This week we pulled live Search Console and GA4 numbers for an acquisition audit. One line stopped the "just ship more SEO pages" conversation:
| Channel (GA4, 28d) | Active users | Sessions | Conversions |
|---|---|---|---|
| AI Assistant (mostly ChatGPT) | 38 | 52 | 5 |
| Organic Search | 22 | 56 | 2 |
Same product. Same month. Different discovery surface.
What did ChatGPT actually open?
Session source was almost entirely chatgpt.com. Top landings:
/en— English product page/— primary landing- A few product-adjacent URLs (
/for/tutors-in-ukraine, templates, alternative/blog pages)
That pattern matters. Models were not dumping people on random 2,000-word posts. They were sending them to short, factual product explanations. If those pages do not contain a citable paragraph, the model invents one from weaker sources — or cites a competitor.
What about Google?
Google is not "dead." It is lagging our rebrand.
In the same 28 days, Search Console showed roughly:
- obricrm.com: ~153 impressions, 14 clicks, average position ~53
- studentportalua.com (legacy, 301 → obricrm): ~1,470 impressions, 21 clicks, average position ~12.5
Path-preserving 301s work. Change of Address was filed. Equity transfer still takes months. Meanwhile, Superprof-alternative queries already show on the legacy property with positions around 8 — volume we want fully credited to the new host.
We also have 432 URLs in the live sitemap. Content volume is ahead of demand. Shipping package after package of compare pages will not fix a domain settling curve or a weak extractable answer on /en.
What we changed because of this (not theory)
Based on those numbers, we prioritized ChatGPT/GEO over classic volume SEO, and paused free→paid product CRO for now.
Concrete moves already in flight:
- Visible "In short" block on the main landings (
/and/en) — one paragraph a model can quote without reading the hero animation. - Stronger FAQ schema on
/en— including "What is OBRI?" and how we differ from Superprof / Preply / italki (marketplace vs CRM). - Sharper Superprof alternatives page — that cluster already had the strongest legacy impressions.
- Updated llms.txt — map Superprof, templates, country pages, and this article for crawlers that read the file.
- Freeze on net-new thin programmatic pages for a few weeks — upgrade winners and LLM surfaces instead.
None of this invents "47 interviews" or fake uplift. It is a response to a small but clear measurement window.
What I am not claiming
- Five conversions is not a growth story for investors. It is a direction signal.
- ChatGPT will not replace Google for every niche forever.
llms.txtis not a ranking cheat code.- Domain migration pain does not mean the 301s failed.
I am claiming this: for a niche B2B tool like a tutor CRM, one clear product page that models can cite currently beats ten thin "alternative" clones that nobody quotes.
What we will do next
- Keep improving extractability on
/en,/, and top alternative/template URLs ChatGPT already touches. - Accelerate indexing and internal links for pages that already rank on the legacy host (Superprof alternatives, progress-report template, Ukrainian payment posts).
- Publish founder notes like this when we have real numbers — then optionally repost to Medium 2–3 weeks later with
rel=canonicalback to obricrm.com. - Leave monetization UX alone until the acquisition mix is clearer (separate workstream).
If you teach privately and a model just recommended "a CRM for tutors," start on the English product page or compare us honestly to a marketplace you already use — for example Superprof alternatives.
That is why we built OBRI: not to win a keyword density contest, but so the next time ChatGPT answers a tutor, the facts it repeats are ours — and true.