News · Fact Check
Was the claim that the Google Trends spikes had odds of less than one in a billion correct?
A biostatistician who studies Google Trends addressed this directly and said the odds calculation is wrong. It treats noisy, sampled data as if each point were an independent fair coin flip.
By Americans for Transparency · Published
The short answer
No. A University of Sydney biostatistician who researches Google Trends addressed the claim directly and said it is incorrect. An odds figure like one in a billion assumes each data point is independent and random in a simple way. Trends data is neither: it is sampled, scaled and carries added noise, and points are correlated. Treating it as a string of fair coin flips produces a dramatic number that the underlying data does not support.
- The one-in-a-billion figure comes from treating Trends points as independent random events.
- A biostatistician who studies Trends said that specific claim is incorrect.
- Trends data is sampled, scaled and noisy, so the independence assumption does not hold.
- Correct analysis requires aggregating many samples, not a probability built on one chart.
The claim, and what the record shows
Claim 1 · A widely shared post about the Charlie Kirk case
The odds of the Google Trends spikes appearing by chance are less than one in a billion, so they must reflect real foreknowledge.
Not supportedA University of Sydney biostatistician who researches Google Trends addressed this post and said its claim is incorrect. The calculation assumes each point is an independent random event, but Trends data is sampled, scaled and noisy, with correlated points. The independence assumption fails, so the probability figure does not mean what the post says.[1][2][3]
Where the number comes from
The one-in-a-billion style figure is produced by treating each spike as an independent coin flip: if the chance of a spike on any given day is small, then several spikes landing on meaningful days multiplies into a tiny probability. That reasoning is only valid if the events really are independent and the per-event odds are known.[1]
Why it does not hold for Trends
Trends data breaks the assumption. It is a sample, not a full count. It is scaled so the peak is always 100. Google adds noise. And neighbouring points are correlated, not independent. On rare terms, as AFNPT's controls showed, spikes appear routinely with no real interest behind them, so the real per-event probability is nothing like the figure the calculation assumes.[2][3]
The biostatistician Jacques Raubenheimer, who studies Google Trends at the University of Sydney, examined a post making this kind of claim about the Kirk case and said its odds figure is incorrect. He added that getting accurate results requires aggregating many samples rather than reading one chart.[1]
“To get accurate results, one has to aggregate many samples of Google Trends data”
What a correct analysis looks like
Instead of one chart and a probability, the sound approach pulls the same query many times and averages the samples, which cancels the noise and reveals whatever real signal exists. AFNPT's control test, running about 1,600 searches, is an example: it showed that spikes on rare terms are common and tell you little on their own.[3]
Questions people ask
Does this mean the spikes definitely had no meaning?
No. It means the probability argument offered as proof does not work. The spikes might have mundane causes or none; the one-in-a-billion figure simply cannot establish foreknowledge.
Who is the biostatistician?
Jacques Raubenheimer of the University of Sydney, who researches Google Trends and wrote about these claims in The Conversation. He said the one-in-a-billion type claim is incorrect.
Sources
- [1]The Conversation (Jacques Raubenheimer, University of Sydney), How misreading Google Trends is fuelling Bondi attack conspiracy theories, December 17, 2025. News report“To get accurate results, one has to aggregate many samples of Google Trends data”
- [2]Google, FAQ about Google Trends data. Official source“These apparent one-off spikes should not be interpreted as real search activity or evidence of individual searches.”
- [3]Americans for Transparency, The Fort Huachuca guest list, checked, October 7, 2026. AFNPT review