Search Engine Journal Reports On AI Visibility Data

Search Engine Journal reported on four datasets that show AI assistants send a small portion of the traffic Google sends. These datasets also indicate that the hidden AI influence many assumed was present in Direct and branded search has not been found. According to Search Engine Journal, this does not mean AI visibility is a sham, but rather that current measurement methods may not pass a basic data quality test.

Recently, Rand Fishkin and Amanda Natividad, co-authors of the book Zero Click Marketing, published warnings about AI visibility, sparking a conversation about its impact. However, Search Engine Journal notes that the data underlying their warnings came from other sources and has been accumulating over time.

Kevin Indig's Growth Memo published a study on October 19, 2026, which found no reliable link between AI visibility and later direct traffic or branded search. Similarly, Marie Haynes' study found that ChatGPT sent 0.62% of Google organic's volume, while all LLMs sent 0.73%. SEO Works' data also showed that organic sent 39 times more traffic than AI assistants.

The traffic numbers from AI assistants are small, and the conversion story is mixed. Marie Haynes' study found that ChatGPT visitors converted at 13.8%, compared to 3.02% for Google organic. However, Google still drove 96.5% of total conversions, and Google organic drove 27 times more conversions than all LLMs combined.

The hidden influence of AI on Direct traffic and branded search has not been found. Kevin Indig's study found that when he filtered for engaged sessions, total Direct traffic rose 140.9%, while engaged Direct changed by -1.0%. This suggests that bots, not buyers, may be driving the traffic.

Amanda Natividad tied this back to an Ahrefs study, which found that branded search volume correlated only 0.392 with AI Overview brand visibility. This suggests that if AI visibility reliably produced branded search, the correlation should be much higher.

Neither Rand Fishkin nor Amanda Natividad concluded that AI does not matter. Instead, they argued that AI is probably influential on some buyers in some sectors and should be invested in and measured as a brand channel. Fishkin credited Mike King with the better frame, that AI is 'more like a brand channel, a digital billboard if you will, and it should be invested-in and measured as such.'

The measurement industry saw this coming in May, when the International Association for the Measurement and Evaluation of Communication launched its GEO Principles. The principles propose four questions that any data set should pass before anyone acts on it, including whether the data is from the sources expected, whether the metadata is complete and faithful, whether the data corresponds to real-world behavior, and whether the data is trusted by those who matter most.

The AMEC Data Quality Principles white paper warns that metrics can be 'internally consistent but externally untrue' and that this kind of data 'does not stay quarantined; it propagates.' This is a precise description of most AI visibility scores, which are consistent from week to week but may not predict anything a business cares about.

The paper also notes that evaluations performed once and stored on a shelf are 'decorative, not protective.' Instead, outputs should be tested 'across tools, prompts, and time.' This suggests that AI visibility scores should be continually tested and validated to ensure they are meaningful and useful.