# The Google AI Overviews lawsuits, and what a tiny site can learn > Publishers are suing over lost traffic. The legal question is not yours to win, but the structural lesson underneath it applies to any site with one channel. Published 2026-09-05 · 5 min read · google-ai-overviews-lawsuits, publishers, traffic Canonical: https://bidsurvivor.space/blog/google-ai-overviews-lawsuits-lesson-for-small-sites --- The Google AI Overviews lawsuits are the loudest media story of the year: publishers arguing that summarising their work at the top of a results page takes the value of the reporting while leaving them the cost of producing it. This is not a legal analysis — I am not qualified to write one, and the outcome will take years. It is about the structural lesson underneath, which does not depend on who wins. ## The shape of the complaint Stripped of legal framing, the publishers' position is: we invested in producing something on the understanding that the traffic it generated would pay for it, and the terms of that arrangement changed unilaterally, without notice, in a way that removed most of the return while leaving the investment intact. The scale is measurable rather than rhetorical. [Search Engine Land](https://searchengineland.com/google-ai-overviews-cut-search-clicks-report-471497) reports search clicks down roughly 42% from baseline and a 61% click-through drop on queries where an Overview appears. [Press Gazette's coverage](https://pressgazette.co.uk/publishers/search-isnt-dead-its-fragmenting-how-to-manage-google-traffic-decline/) describes the effect on newsrooms specifically. Whether that constitutes a legal wrong is genuinely contested and genuinely uncertain. What is not contested is the mechanism: an intermediary changed its behaviour, and everybody downstream absorbed the difference. ## The lesson has nothing to do with Google If you run a small site, here is the part that transfers, and it is uncomfortable: **the publishers did nothing wrong that you are not also doing.** They built on the best available channel, they built well, and the channel's terms changed. Their mistake, insofar as there was one, was structural rather than tactical — the return on a large, slow, irreversible investment depended entirely on a counterparty who had never promised anything. That is the same bet as: a business whose customers all arrive from one search query, one app store's featured list, one social platform's recommendation, or one marketplace's category page. The size differs. The exposure does not. Note what the lesson is *not*. It is not "do not use big channels" — that is unaffordable advice for a small site, and the big channels are where the people are. It is narrower: **do not let the payback period of your investment exceed the notice period of your channel.** Publishers were writing articles whose economics assumed years of search traffic, on a platform that could and did change in a quarter. The mismatch between those two horizons is the whole problem. ## What that implies concretely **Prefer investments that pay back inside a quarter.** Not because short-term work is better, but because a shorter payback survives a channel change. This is the same argument as the one for [advertising that expires](/blog/advertising-that-expires): a thing with a clock on it cannot be devalued by a rule change three years from now, because it will already have been paid for. **Own the endpoint even when you rent the channel.** A visitor who arrives from anywhere and leaves an email address has converted a rented relationship into an owned one. That is the single highest-leverage move available, and it is unglamorous enough that most sites under-invest in it. **Assume the terms will change and ask what breaks.** Not as a doom exercise — as a specific question with a specific answer. If your largest channel halved tomorrow, what would you do in week one? If the answer is "start building the second channel", start now, because week one is the worst possible time. **Prefer channels whose economics you can verify.** A channel that reports its own performance privately, after the spend, gives you no way to notice a change until it has already cost you. One where the numbers are visible independently at least lets you see the shift as it happens. That is a real if partial defence, and it is the argument behind [platforms that publish their numbers](/blog/ad-platforms-that-publish-their-numbers). ## Where the analogy breaks Two honest limits. Publishers had scale and contracts and legal budgets and still got caught, which means "be more careful" is not a complete answer and it would be glib to pretend otherwise. Some of this is not manageable at any size; it is a structural feature of building on infrastructure somebody else owns. And a small site has one genuine advantage the publishers did not: **you can change quickly**. A newsroom with three hundred staff and a decade of archive optimised for one distribution model cannot pivot in a quarter. A one-person site can test a new channel this week. The correct use of that advantage is not to feel superior about it — it is to actually exercise it, regularly, before you need to, which is the practical reason for the [two channel rule](/blog/the-two-channel-rule). ## What I would not conclude I would not conclude that search is finished, that AI companies are villains, or that anybody should stop writing. The traffic that remains is still substantial, and a lot of the commentary is people with something to sell attaching themselves to a large story. The defensible conclusion is smaller and more useful: **an intermediary with no obligation to you will eventually act in its own interest, and the only real protection is a payback period short enough that it does not matter when they do.** That was true before this, it will be true after whatever the courts decide, and it is one of the few pieces of strategy that does not expire.