ChatGPT sends about 0.62% of the organic traffic that Google Search sends, according to an analysis by Greg Jarboe published in Search Engine Journal. Jarboe's argument is that AI visibility is real, but brands can only defend spending on it when OpenAI's ChatGPT and similar tools are measured by their effect on customers. That means checking dashboard scores against customer behavior and tracking exposure, recall and consideration.
- Who: Greg Jarboe, writing for Search Engine Journal.
- What: An analysis that draws on four datasets about AI visibility (Search Engine Journal).
- Headline figure: ChatGPT sends 0.62% of Google's organic traffic (Search Engine Journal).
- Recommendation: Test AI visibility dashboard scores against customer behavior, and measure exposure, recall and consideration (Search Engine Journal).
What does the 0.62% figure mean for AI visibility?
Put simply, for every 1,000 organic visits Google sends to a site, ChatGPT sends roughly six. That ratio should change how marketing teams talk about AI search. ChatGPT gets heavy coverage in the trade press and in boardrooms. As a source of clicks, though, it is still small compared with Google.
The figure also fits a broader pattern. Whizsky has reported that OpenAI itself puts Google's search volume at about five times ChatGPT's. A gap in query volume is one thing. The gap in referral traffic appears to be much wider, and the reason is how the two products work.
Google Search is built to send people to other websites. ChatGPT is built to answer the question inside the conversation. A brand can be mentioned, recommended or compared in a ChatGPT answer and never get a visit. That is the core measurement problem with AI visibility. If clicks are the only metric, much of the influence goes uncounted. If influence is counted with no link to outcomes, budgets rest on numbers that cannot be checked.
Why most AI visibility dashboards are hard to defend
Many SEO and analytics vendors now sell AI visibility scores. Most of these products run a set of prompts through large language models and record how often a brand appears in the answers. That data has value. Competitive tracking in AI answers, which Whizsky covered in its guide on how to track competitors in ChatGPT answers, can show gaps in positioning and content.
These scores have three common weaknesses when they are used to justify budgets:
- Prompt sets are artificial. A vendor's prompts may not match the questions real buyers ask. The phrasing may differ, the topic may differ, or the stage of the purchase journey may differ.
- AI answers vary from run to run. LLM responses can change between sessions, users and model versions. A single snapshot can make a score look more precise than it is.
- Appearing is not the same as influencing. A mention in an answer does not show that anyone read it, remembered it or acted on it.
Jarboe's framing deals with all three. He does not throw out the dashboards. He recommends testing them: if a visibility score goes up, some measurable change in customer behavior should follow.
How to measure AI visibility: exposure, recall and consideration
The three metrics Jarboe highlights are exposure, recall and consideration. They come from classic brand measurement, and they apply well to AI search because they do not depend on clicks.
Exposure
Exposure asks whether real people actually saw the brand in an AI answer. Visibility trackers give a proxy for this. Teams can improve that proxy by building prompt sets from their own data, such as sales call notes, site search logs, support tickets and Google Search Console queries. Prompts that look like real buyer questions give an exposure number that is easier to trust.
Recall
Recall asks whether people remember the brand afterwards. Brand lift surveys, self-reported attribution fields and post-purchase questions can capture it. Adding a "How did you hear about us?" field that lists ChatGPT and other AI assistants is cheap and gives direct evidence.
Consideration
Consideration asks whether AI exposure moves the brand onto the shortlist. Useful signals include changes in branded search volume, direct traffic, demo requests and mentions of AI tools in sales conversations. If AI visibility is working, some of its effect may show up as branded Google searches instead of ChatGPT referrals.
| Metric | What it answers | Practical signals |
|---|---|---|
| Exposure | Did buyers see the brand in AI answers? | Visibility tracker scores built on real customer prompts |
| Recall | Do buyers remember the brand? | Brand lift surveys, self-reported attribution |
| Consideration | Is the brand on the shortlist? | Branded search trends, direct traffic, sales call mentions |
How should marketers budget for AI search in 2027 planning?
A small traffic share does not mean a brand should ignore AI search. It means spending should be sized to the evidence. Teams going into 2027 planning can use these steps:
- Protect Google first. Google is still the main source of organic traffic, so core SEO work should keep most of the investment.
- Run AI visibility as a test. Set a defined budget with a set time frame and agreed metrics, rather than an open-ended line item.
- Validate vendor scores. Check whether changes in a dashboard score line up with changes in branded search, surveys or pipeline. If they never do, question the tool before increasing spend.
- Watch the paid side. As OpenAI builds out advertising, including the $3–$5 CPC bidding model Whizsky reported on, paid placements will produce click data that organic AI visibility cannot. Comparing the two could show how much an AI mention is worth.
Teams that adopt this discipline early are better placed whichever way AI search develops. If ChatGPT's referral share grows, they will already have a baseline. If it stays small, they will not have overspent on a channel based on numbers no one could verify.
Key Takeaways
- According to Search Engine Journal, ChatGPT sends 0.62% of Google's organic traffic, so Google is still the dominant traffic source.
- AI visibility matters, but click-based metrics miss much of its influence, and dashboard scores alone cannot prove that influence.
- Defensible AI visibility budgets come from testing vendor scores against real customer behavior.
- Exposure, recall and consideration give marketers a framework that works even when AI answers produce no click.
- Run AI visibility spend as a measured test, and keep it a modest share of the budget until outcome data supports more.
Frequently Asked Questions
How much traffic does ChatGPT send compared with Google?
According to an analysis by Greg Jarboe in Search Engine Journal, ChatGPT sends about 0.62% of the organic traffic Google sends. That works out to roughly six ChatGPT visits for every 1,000 Google organic visits.
Is AI visibility worth investing in if ChatGPT traffic is so low?
It can be, because AI answers can influence buyers without sending a click. Spending should be sized to the evidence, though. Brands should check whether AI visibility improves recall and consideration before committing large budgets.
How can marketers measure AI visibility accurately?
Search Engine Journal recommends testing dashboard scores against customer behavior and tracking exposure, recall and consideration. In practice that means building prompt sets from real buyer questions, running brand lift or attribution surveys, and watching branded search and sales signals.
Why are AI visibility dashboard scores unreliable on their own?
They often depend on artificial prompt sets, and LLM answers vary between sessions and model versions. A brand mention in an answer also does not show that anyone saw it, remembered it or acted on it.
