News · Fact Check
Does running the same Google Trends search twice give you the same chart?
Not on rare terms. Google draws a fresh random sample for each request, so the same query on the same settings can return different dates and spikes hours apart.
By Americans for Transparency · Published
The short answer
Not reliably. Google Trends draws a new random sample each time a query runs, so the same search with the same settings can return a noticeably different chart from one request to the next. On popular terms the differences are small. On rare terms the spikes can move to different dates or vanish. A chart that cannot be reproduced on demand is not a fixed record of what happened.
- Google Trends samples the data anew for each request rather than serving a fixed record.
- AAP FactCheck documented the same query producing noticeably different charts on different days.
- AFNPT ran the Fort Huachuca searches hours apart and got different dates for the spikes.
- Reproducibility is a basic test of evidence; these rare-term charts fail it.
The claim, and what the record shows
Claim 1 · Implied by posts presenting the charts as proof
A Google Trends screenshot is a fixed record, so the spike it shows will be there whenever anyone checks.
Not supportedGoogle draws a new sample for each request. AAP FactCheck and AFNPT both found the same rare-term query returning different charts on different runs, with spikes moving dates or disappearing. A reading that changes between runs cannot serve as a fixed record of a past event.[1][2][3]
Why the chart moves
Google Trends does not store a single, canonical chart for each query. When you ask for a term, it samples the underlying search logs and builds the chart from that sample. Ask again and it samples again. On terms with lots of volume the samples look almost identical. On rare terms, where a handful of sampled searches drives the whole line, the samples can differ a lot.[1]
What the checks found
AAP FactCheck tested this directly and reported that running the same search on a different day can produce a noticeably different result, which is why a single screenshot proves little on an uncommon term.[2]
AFNPT ran the Fort Huachuca searches at different times on the same day with identical settings. The spikes landed on different dates between runs. That instability is the expected behaviour of a resampled index, and it is fatal to the idea that any one screenshot pins down a real event. The test record is here.[3]
Why reproducibility is the point
Evidence that cannot be reproduced on demand is weak evidence. If the same query, run twice, gives you two different answers, neither answer can be treated as a measurement of what happened. The way to extract a real signal, researchers say, is to average many samples, not to trust the one that was screenshotted.[2]
Questions people ask
Does this happen with popular search terms too?
Technically yes, but invisibly. With high volume the samples are nearly identical, so the chart looks stable. The instability only becomes obvious on rare terms with little real search volume.
Could I reproduce the Kirk spikes myself?
You might see a spike, but not necessarily on the same date, and not every time. AFNPT and AAP both found the results shifting between runs, which is the opposite of a reliable record.
Sources
- [1]Google, FAQ about Google Trends data. Official source“This means that for these queries, Trends data will sometimes reflect statistical noise rather than actual search interest.”
- [2]AAP FactCheck, Bondi attacker search data is static, not a conspiracy, December 24, 2025. News report“Similar claims involving the results from the website were made after US conservative Charlie Kirk's assassination.”
- [3]Americans for Transparency, The Fort Huachuca guest list, checked, October 7, 2026. AFNPT review