A court filing in the consolidated copyright case against OpenAI and Microsoft has surfaced internal messages that ought to settle one of the more comfortable assumptions still circulating in search marketing: that being cited by an AI system is functionally equivalent to being visible in search results. Search Engine Journal reports that the publishers' brief, filed 17 September in federal court in New York, quotes an OpenAI software engineer saying "no matter how prominently we show the links, users won't click." The New York Times, one of the plaintiffs, reported the engineer wrote that in February 2023.
The same filing quotes Nick Turley, OpenAI's head of ChatGPT, on the product's browse function. According to the brief, Turley said there was "no good reason to click" on a link to the original source once the chatbot had provided an answer. These are arguments in one side's case, not established facts, and the court has not ruled. But they align closely with data Microsoft itself provided, and with what the citation and impression metrics platforms now publish actually measure.
The Microsoft click-through data
The publishers' brief cites Microsoft data comparing click-through rates for the plaintiffs' websites in Bing Chat and Bing Web Search. For The New York Times's sites, CTR was 87% to 93% lower in Bing Chat. For the Daily News plaintiffs' sites, 83% to 91% lower. For Ziff Davis sites, 51% to 94% lower.
The filing defines CTR as the percentage of clicks to a URL relative to its impressions, described as appearances of that URL in a search results page or a chatbot response. The ranges do not measure traffic or visits. The passage does not state when Microsoft collected the data, how many queries or impressions it covered, or how the sample was selected. It also does not explain what the low and high ends of each range represent. Microsoft renamed Bing Chat to Copilot in November 2023, but the product name does not date the figures, and nothing in the brief ties them to the current version of Copilot.
The publishers argue the reductions "clearly show substitution." The same passage notes they are not required to prove harm with data and that "common sense suffices." OpenAI and Microsoft dispute the case, with both filing summary judgment motions on 4 September arguing their use of news content is fair use. Axios reported that OpenAI's motion cited studies suggesting ChatGPT outputs do not create commercial substitutes for The Times's work, and Microsoft made similar claims about Copilot.
Why citation is not traffic
The practical problem for most marketing teams is that platform reporting now confirms the gap the Microsoft data illustrates. Microsoft's AI Performance report in Bing Webmaster Tools counts citations. The help documentation states explicitly that citation metrics differ from traditional search metrics that measure rankings, clicks and traffic. When Microsoft added Citation Share in June, the announcement noted the metric does not represent traffic share.
On Google's side, Search Console's generative AI report shows impressions. It does not list clicks. The metrics platforms are now publishing measure appearance, not the behaviour that follows. The Microsoft data in the publishers' filing covers that next step: how often links to three publisher groups' sites were clicked in Bing Chat answers compared with how often they were clicked in Bing search results.
When Microsoft added chat data to the Webmaster Tools Performance report in 2023, it combined web and chat clicks and impressions in one view. The CTR figures in the court filing separate the two. That separation is what makes the data legible. A blended number masks the structural difference between how people use a search results page and how they use a conversational interface that is designed to answer the question without requiring an onward click.
What this means for AI visibility work
If your AI visibility measurement still treats citation and traffic as roughly the same thing, these figures should change that assumption. The engineer's message and Turley's comment describe user behaviour that the Microsoft CTR data supports. Being cited is a prerequisite to being clicked, but it is not a proxy for it, and the gap between the two appears to be structural rather than temporary.
The immediate operational question is whether your reporting separates citations from clicks, and whether the targets and success definitions for AI visibility work account for that gap. If the brief is correct that Bing Chat CTR for a major news brand was 87% to 93% lower than Bing Web Search CTR, that is not a rounding error or a product quirk to wait out. It is the central design characteristic of the format.
There are strategic implications as well. Citation alone may still have value: brand visibility, authority signalling, defensive share of voice. But those outcomes need to be measured and valued separately from the traffic and conversion outcomes that search visibility has historically delivered. The work required to earn a citation is different from the work required to earn a click from a search result, and the return on that work is different too.
The practical audit question
In AI visibility audits, the first diagnostic step is understanding what the current measurement actually tracks. If the platform reports citations or impressions, those numbers do not tell you whether anyone clicked. If internal dashboards blend AI and traditional search clicks, the aggregate number hides whether the channel mix is shifting and whether performance in one channel is masking decline in another.
The second step is deciding what citation is worth in the absence of a click. That depends on the business model, the audience and the goals of the content. For some organisations, being the answer is enough. For others, particularly those whose business depends on traffic to a monetised site, citation without clicks is not a sustainable outcome. The measurement has to reflect that reality, or the strategy built on it will optimise for the wrong result.
What happens next in the case
Judge Sidney H. Stein's 24 September order set 23 October for oppositions, 30 October for amicus briefs, and 20 November for replies. Whether OpenAI and Microsoft respond to the engineer's message, Turley's comment or the CTR data in their October filings will be worth watching, as will any additional context for how Microsoft measured the figures or when the data was collected.
For marketing and search teams, the case is less important than the data. The quotes are one side's argument in litigation. The CTR comparison is descriptive, and it describes a pattern that matches what the platforms' own reporting now separates into different metrics. If you are still treating AI citations as a like-for-like replacement for search visibility, the Microsoft figures should prompt a closer look at what your own data shows when you split the two.