Google integrates AI-generated preview data into Search Console. For SMBs, this redefines reading organic traffic and forces a rethink of content strategy.
For several weeks, Google has been enriching its Search Console with new metrics related toartificial intelligence. For SMEs, understanding what this data really means could change the way they manage their SEO.
A finally honest dashboard
For months, webmasters had been observing a worrying anomaly: stable positions, even increasing, but crumbling traffic. The reason is now better documented. The integration of AI-generated previews into Google results captures a growing share of Internet users’ attention without generating clicks to the source sites. Search Console is, finally, beginning to expose this reality.
The new data columns distinguish impressions from traditional organic results from those produced as part of generated responses. This is considerable progress. Until now, a marketing manager or an SME manager could not know if their result sheet had been used to feed an automatic response or if it had been clicked by a human. This opacity maintained an illusion of good SEO health that the conversion figures belied.
Google didn’t reveal everything at once. The data remains fragmented, the filters imperfect. But the signal is there, and it is useful: for the first time, we can begin to distinguish what generates visibility from what generates real traffic.
Control what AI consumes
The other new feature, less commented on but just as structuring, concerns the control options granted to publishers. Via Search Console and via the robots.txt file, it is now possible to modulate the access of AI robots to the content of a site. Google-Extended, the crawler used to train some Gemini models, can be blocked independently of the traditional Googlebot. Other third-party robots, such as GPTBot from OpenAI or ClaudeBot from Anthropic, are affected by the same levers.
This is not trivial. A publisher can now decide to remain visible in traditional search results while refusing to allow its content to be used as raw material for competing systems. This granularity was non-existent eighteen months ago. It represents a form of editorial sovereignty that many have demanded.
There remains the question of real effectiveness. Blocking a well-identified crawler does not prevent other less transparent agents from passing through. And in fact, the content that ends up in Google’s AI previews is not necessarily that which the training robot ingested: the mechanisms are distinct. We must therefore separate two often confused problems: the question of training models on the one hand, and that of quoting in the generated responses on the other.
A concrete challenge for SMEs
For a large company with an SEO team, these new variables fit into existing processes. For an SME, it’s a different story. The business manager who manages his own online presence does not have the time to decode dashboards whose logic changes every quarter. And yet, the stakes are real: if his product sheet feeds the AI responses without generating visits, he pays the cost of content production without receiving the profits.
Leaders need to know if their content is working for them or someone else. The answer isn’t in average click-through rate: it’s in cross-reading the new Search Console columns with actual conversion data. This work is tedious to do manually. However, it becomes essential as the share of answers generated by AI increases in the results pages.
The good news is that SMEs that produce expert, local and specific content often have a natural advantage in this new context. Generic AI generalizes. The SME which documents its real job, its real constraints, its real answers to specific questions, produces content that models have difficulty reproducing in a credible manner.
Rethinking what it means to be visible
The “position in Google” KPI is losing part of its value as a performance indicator. This is not the end of SEO, it is a transformation of what it measures. Being first on a keyword whose results are 80% covered by an AI overview is like having the best window on a street whose traffic has changed without anyone having announced it.
Relevant visibility is defined differently today. It takes into account the actual click rate, the nature of the queries on which the site appears, and the proportion of organic clicks in the overall acquisition mix. New Search Console tools provide answers to these questions. They don’t give all the answers yet, but they finally pose the right problems.
In conclusion
The Search Console update is not a minor technical event. It marks a step in the official recognition, by Google, of a sharing of value which has not yet found its balance. For SMEs, the issue is not to fear AI or block it on principle, but to learn to read what it actually does with their content. The tools are improving, the data is becoming more readable, the control levers exist. The fundamental question remains open: how to build an online presence that remains profitable when the response arrives before the click?
