Given Openai’s close relationship with Microsoft, many people expect Chatgpt’s Retrieval Generation (RAG) process to use Bing.
However, in the past few months, Some SEO Discover The chat may have been secretly turned to Google.
These are all one-time experiments. So, I wonder if it is possible for us to find the “truth” using data.
This is what we found.
I asked our data scientist xibeijia guan Help with this. Here’s what she did:
- She extracts the actual search query ChatGpt (“Fan Exit Query”) and returns the URL from those searches. These data come from us Ahrefs brand radar.
- She then ran the exact same search query through Google to see the URL Google would return
- She measured the frequency of URLs returned by Chatgpt appearing in the top 10, top 20 and anywhere in Google’s search results.
On average, Chatgpt extracts 1.78 search queries per prompt, with 75% of the prompts triggering exactly two searches.
average:
- The only one 6.82% Chatgpt search results are in the top ten Google SERP
- The only one 9.85% Top 20 Chatgpt search results
- The only one 16.61% ChatGPT search results are in Google’s SERP.
If Chatgpt is just taken from Google’s search results, you’ll expect higher overlaps. instead, 83.39% The results selected by Chatgpt do not appear in Google’s search results at all for the same fans to be out of query.
My colleague Louise also looked at 15,000 tips and found this average Only 12% of links Quote by Chatgpt, Gemini and Copilot, appears in Google’s top 10 results, the same prompt.
Therefore, there is no clear indication that chatgpt is only or Mainly Use Google as a search engine.
Chatgpt might use a hybrid approach that can get search results from various sources, such as Google SERP, Bing SERP, their own indexes and third-party search APIs, then combine all URLs and apply their own rearrangement algorithms.
The final thought
Chatgpt doesn’t seem to be “secretly driven by Google.” Instead, it seems to use a complex multi-source approach.
From a product perspective, this makes sense.
OpenAI may want to reduce dependency on any single search provider while optimizing its specific use cases: provide accurate contextual answers, rather than general web discovery.
Have any questions or comments? Let me know LinkedIn.
