One experiment a week. Free while we build the archive.

GetCited Labs runs controlled experiments on what ChatGPT, Claude, Perplexity and Google AI Overviews actually cite — on our own estate of sites, because the only properties you can run a destructive test on are ones nobody depends on.

We publish one every week: the hypothesis, the pre-registered pre-state, the raw data, the confounds we killed, and the verdict — including when the verdict is that nothing happened.

What a member gets

• Each experiment as it publishes, before it circulates
• The full archive, and the raw facts.json behind every result — the same file we write before the prose
• The queue: what we're testing next, and the ability to propose an experiment

What it costs

Nothing yet. The archive is four experiments deep, which is not enough to charge for, so we don't. The founding cohort is free until the archive reaches ten.

After that it becomes paid. Founding members are not charged during the founding period and will be offered a permanent founding rate — we'll tell you the number before it applies, and leaving is one click. We'd rather under-promise here than have you find out later.

What you are not buying

Findings. Members never influence what we test, how we measure it, or what the result says. We don't sell placement in an experiment, sponsored findings, or a favourable reading, and no commissioned work will ever change a published result. Every experiment stays free to read — the membership pays for access and the data, never for conclusions. The methodology page states both rules in full, along with the properties we operate in markets we measure.