Why clicks are the wrong metric for AI visibility
The click misunderstanding
In classic Google search, many searches lead to a click. We have gotten used to that for 25 years. AI answers work differently: the answer is the result. Often there is no reason to click anywhere.
The numbers
In a Google search, roughly 30 percent of searches lead to a click. In Google's AI mode it is more like 5 percent, in ChatGPT on search-like questions around 1 percent. Counting only the incoming clicks makes ChatGPT look tiny when it is not.
Dark search: impact without a trace
Decisions increasingly happen directly in chat, invisible to classic analytics. This resembles word of mouth: a recommendation works but leaves no measurable click path. In B2B, companies already report that 20 to 30 percent of their leads come from AI recommendations, according to direct surveys, not analytics.
The consequence
AI visibility belongs to brand-marketing thinking more than to performance marketing. Instead of clicks, mentions, citations and share of answers count. Ignoring this means steering by a metric that does not capture the actual effect.
Key takeaways
- AI answers produce hardly any clicks (ChatGPT around 1 percent on search-like questions).
- Click-based measurement dramatically underestimates the channel.
- The impact is often dark: real, but without a click trace.
- Useful measures: mentions, citations, Share of AI.
Frequently asked questions
Does that mean clicks do not matter for AI visibility?
No, they remain an additional signal. But as the main measure for AI visibility they do not work.
How do you measure AI visibility instead?
Through mentions and citations in prompt monitoring as well as self-reported attribution.
Where does the 1 percent ChatGPT click-through rate come from?
From industry observations on search-like prompts in ChatGPT. The exact figure varies, the order of magnitude is small.