Who they are and what each did

Baron Coleman is a Montgomery, Alabama attorney and broadcaster. On a stream published September 15, 2025, five days after Charlie Kirk was shot, he ran Google Trends searches on air for terms tied to the case. A 16-minute cut of that stream was republished on September 24, 2026.^[1][2][3]

Ana Escobar's work is different in subject and in claim. She searches the names of officials in the television market around Fort Huachuca, Arizona, and reads the spikes as a list of people who visited the base. AFNPT reviewed that claim in full.^[4]

What Coleman showed

How the two methods differ

The first row is the one that matters most. Coleman presents the charts as leads to be checked. Escobar presents them as a guest list. A chart of sampled, scaled search interest can support the first use far better than the second.^[3][4]

Coleman also explained why he cut the days after the shooting out of his date ranges. Every term tied to the case shot up after September 10, and he wanted to know whether he was looking at statistical noise around the event. For the tail number he went back 13 months in the Washington, D.C. area and reported no interest until about a week and a half before the shooting.^[3]

The record is not one-sided on Escobar. AFNPT's review notes that she described Trends as a sampling method herself, that she warned a plain common name proves nothing, and that her date range also leaves out the post-shooting surge. Where the two part ways is in what they say the spikes show. Where she has a point.^[4]

When AFNPT ran control terms with her settings, invented and cartoon names produced spikes on her dates. When her own searches were re-run, none of the 31 terms with a known date came back within a day of it.^[4]

Why his results are still inconclusive

Coleman's checks answer one objection: that the spikes are just the echo of the news. They do not answer the larger one. On a term that almost nobody searches, Google says a spike can appear with no real search behind it.^[6]

Several of his findings are exactly that shape: one point, on a rare term, in one metro area. A long flat baseline does not make such a point more reliable. Rare terms are where the noise shows most.^[6][7]

Looking in many places also matters. Coleman told viewers he "looked a lot of places." The more terms and regions anyone tries, the more isolated points they will find. In AFNPT's tests with Escobar's settings, about one rare or made-up phrase in three showed a spike somewhere. That rate was measured with her settings, not his, so it is a guide and not a measurement of his searches.^[3][4]

The seven-day charts add a further limit. Google says charts of seven days or less use the viewer's own time zone. Coleman broadcasts from Alabama and the shooting was in Utah, so the hours on his screen would be Central time unless his device was set otherwise.^[6]

The moment a result disappeared

Near the end of the segment Coleman re-ran a search and the result he had shown was no longer there.^[3]

He returned to it in a follow-up stream. A friend had called to say the searches could not be reproduced. Coleman said he tried again with the same dates and regions and found that most of them no longer appeared.^[8]

There is a documented explanation that does not involve anyone removing data. AAP FactCheck reported that Google draws a different random sample each time a query is run, and that for uncommon terms the same search on a different day can give a noticeably different chart. AFNPT saw the same thing in its own tests: searches run hours apart returned different dates.^[9][4]

This is the point where a careful method and Google's design collide. A result that cannot be reproduced can be read as proof of tampering or as proof that the first result was noise. The chart cannot tell you which.

What critics have said about Coleman's work

Gary Melton, who publishes as Paramount Tactical, said in a November 2025 video that he entered every one of Coleman's searches with the same settings and never reproduced a result.^[10]

James Li, whose October 2025 video carried Coleman's charts to a wider audience and praised his research, told his own viewers: "Full disclosure, I did try to replicate these search trends, but I could not." Li then relayed claims that the results had been scrubbed, and also read out an explanation that sampling and normalization are overwhelmingly the likely cause.^[11]

A write-up that was otherwise sympathetic to Coleman's segment conceded two points: that nothing ties the searches to government offices or media insiders, and that Google's regional scaling can make a small number of searches look large. Distractify, covering the same charts as shared on TikTok, said it checked Google Trends and did not find the searches, and called the theory unproven.^[12][13]

Jacques Raubenheimer, a University of Sydney biostatistician who studies Google Trends, used the accused man's full name as a worked example of how rare terms produce isolated spikes. He did not name Coleman.^[7]

AFNPT found no report by a wire service or fact-check desk that examines Coleman's charts directly, and no independent reproduction of them, as of October 13, 2026.

Where Coleman's own framing goes further than his caveats

Coleman's caveats are real and frequent. So is a second thread in the same segment. He told viewers he was about to show "the fact I think that could put it all together for people who want to draw a conclusion," and asked what else he would need to search "if I'm plotting an assassination." At one point he referred to "the roommate of the assassin" without the word alleged. Tyler Robinson is charged, has pleaded not guilty and is presumed innocent.^[3][14]

That does not undo the method. It is a reason to separate what he did, which was careful, from where the segment points, which the data cannot support.

The fair summary

Coleman treated the charts as leads and said so. Escobar treats them as a record. On method, that is a large difference, and it is to Coleman's credit. On what the charts can prove, there is no difference. Without the search counts that only Google holds, a pre-event spike on a rare term is consistent with one real search, with several, or with none.