AI’s Impact Is Outrunning Measurement: The Trust And Attribution Gap Facing Brands

Within the first half of 2026, search behavior shifted beneath our toes, software program valuations cratered on sentiment somewhat than proof, and firms blamed layoffs on AI lengthy earlier than anybody might present the receipts.

Kevin Indig posted a hyperlink on LinkedIn final week to his AI Halftime Report, H1 2026, and it’s value noting that his H1 2025 report predicted each Google’s continued AI Mode rollout and the concept AI layoffs had been largely a PR narrative somewhat than an operational actuality. Each predictions held up, and the report opens with a line that doubles because the thesis for your complete first half of the 12 months: AI’s affect saved rising quicker than anybody’s skill to measure it. That hole between affect and measurement is the actual story of H1 2026, greater than any single product launch or earnings name.

Belief Turned A Rating Issue Readers Really feel In Their Personal Outcomes

Indig’s analysis discovered that roughly three out of 4 customers choose the highest lead to an AI shortlist, except a brand they already trust seems wherever else on that checklist, through which case they choose the trusted title as a substitute. Software program shares fell by near 30% over the interval, and the decline tracked nearly solely with how the market perceived an organization’s publicity to AI disruption somewhat than with how that firm really carried out. The underside quartile of software program shares dragged the entire sector down whereas the median and prime quartile outperformed the broader ETF, which tells you the selloff was a narrative about narrative, not fundamentals.

Meta engineers reportedly burned by 73.7 trillion tokens in a single month chasing an inner leaderboard that ranked greater than 85,000 workers by token consumption. No person might level to the return on that spending, and Meta shut the leaderboard down in April as soon as CFOs realized annual token budgets had been blown by in 4 months.

AI Took The Blame For Cuts It Did Not Make

Layoffs instructed an identical story of narrative racing forward of proof. Challenger, Gray & Christmas discovered AI cited as the rationale behind greater than 87,000 job cuts by Could, roughly a fifth of all 2026 layoffs to that time. Indig’s personal reporting, going again to his H1 2025 predictions, argues AI is usually a handy clarification for cuts that had been actually about pandemic over-hiring and capital expenditure self-discipline. A number of the similar firms that blamed AI for layoffs circled and quietly rehired.

My take (and I’ve been saying versions of this since spring) is that Indig has named one thing structural somewhat than seasonal. Each one in all his H1 storylines, the SaaS selloff, the AI layoff narrative, the traffic collapse publishers at the moment are combating in courtroom, comes again to the identical root trigger. We constructed our measurement scoreboards for traditional search years in the past, and none of them had been designed to show one thing as nonlinear and probabilistic as AI search has turned out to be.

That nonlinear narrative exhibits up starkly in Indig’s quotation knowledge. He discovered that 91% of citations seem in solely one in all ChatGPT, Perplexity, or AI Overviews, by no means in a couple of. Prompt tracking, he argues, must behave extra like polling and focus-group analysis than just like the rank monitoring SEOs have leaned on for twenty years. Brand mentions, not citations, correlate extra carefully with actual enterprise outcomes, since most consumers care whether or not a model exhibits up favorably throughout a panel of prompts somewhat than whether or not one particular quotation landed in a single particular reply. Google’s personal Search Console data is reportedly 75% incomplete for this new panorama, so even practitioners who assume they’re measuring fastidiously are working from a partial image.

There was an upside to all that token burning. Indig factors out that the wave of utilization round Claude’s Opus releases pushed way more individuals into daily AI use than any advertising and marketing marketing campaign might have, and that surge fed development at infrastructure firms additional down the stack, the sort of second-order impact that not often exhibits up in a quarterly earnings name however shapes a whole ecosystem anyway.

The agent market itself reshuffled simply as a lot because the metrics did. ChatGPT’s share slid from 78% to 56% between July 2025 and July 2026, whereas Gemini climbed from 15% to 30% and Claude grew from 2% to 10%. Mannequin alternative has grow to be a real enterprise threat somewhat than a choice, which is another variable that resists simple measurement.

Publishers Took The Battle To Courts And Regulators

Publishers spent H1 2026 combating the identical battle in a special area. A Munich court dominated Google responsible for false statements generated by AI Overviews. Four hundred newspapers sued OpenAI and Microsoft over unauthorized content material use. The UK’s Competition and Markets Authority ordered Google to present publishers extra management and transparency over how their content material is utilized in AI search, together with opt-outs from AI Overviews, AI Mode, and Uncover summaries. Each a kind of fights is absolutely an argument over who will get to manage entry and attribution as soon as a human is not the one clicking by.

The Second Half Splits Intelligence From Company

Indig closes his report with a preview of the second half that deserves as a lot consideration as something within the H1 recap. He argues H2 2026 will “separate intelligence from company.” Mannequin functionality retains getting cheaper and extra commoditized by open-weight competitors, however the permission to truly act on somebody’s behalf, spend cash, attain knowledge, or impersonate an individual by an agent, is getting locked down more durable by platforms, governments, cost networks, and customers themselves. The writer lawsuits and the UK CMA order are early proof that this tightening has already began.

That distinction issues extra for search engine optimization technique than most of what acquired consideration in H1. Here’s what practitioners can do with it heading into the second half.

First, retire the idea {that a} single rank-tracking instrument covers AI search. Construct a immediate panel that spans ChatGPT, Claude, AI Overviews, AI Mode, and no less than one open-weight mannequin, and deal with the outcomes the best way a pollster treats a pattern somewhat than the best way an search engine optimization treats a SERP. Given how little overlap Indig discovered between engines, monitoring just one is near monitoring none.

Second, shift reporting away from citation counts and towards point out frequency, sentiment, and advice rank throughout that panel. A dashboard that solely counts citations is measuring the smaller half of what really drives purchaser conduct, and it’ll maintain understating your actual AI footprint to shoppers or management.

Third, begin auditing your agentic access layer now, properly earlier than H2 forces the problem. Test whether or not your product data, pricing, and checkout flow are structured so a licensed AI agent can really full a transaction on a buyer’s behalf, and whether or not your model exhibits up as a trusted, nameable choice when an agent is selecting amongst a number of opponents. Indig’s break up between intelligence and company means the winners in H2 would be the manufacturers brokers are permitted to behave on behalf of, not merely the manufacturers that get talked about most frequently.

I agree with Indig that the power to measure AI’s affect fell behind the affect itself in H1 2026, and I don’t assume that hole closes by itself between now and the top of the 12 months.

Extra Sources:

Featured Picture: Accogliente Design/Shutterstock