--- title: "What I Would Do About a 527 Percent AI-Traffic Jump" description: "AI-referred web sessions rose 527 percent in Previsible's 19-site sample. I would keep SEO and add a citation layer while few brands track AI search." canonical: "https://jaxonparrott.com/blog/the-527-percent-inflection-point" date: "2026-03-26" last-updated: "2026-11-12" --- # What I Would Do About a 527 Percent AI-Traffic Jump I would keep SEO and add a second scoreboard for whether answer engines cite us, because AI-referred web sessions across 19 GA4 properties grew 527 percent year over year, from 17,076 in January to May 2024 to 107,100 in the same months of 2025 ([Previsible's AI Traffic Report, as reported by Search Engine Land](https://searchengineland.com/ai-traffic-up-seo-rewritten-459954)) while most teams are still reporting the old channel as if it were the whole market. The rankings can look fine while the queries that close move into a synthesis we are not in. The citation record I would read next to this demand shift is [AI Citation Gap: Earned Media Drives 82% of AI Citations](https://authoritytech.io/curated/profound-unicorn-citation-gap-wake-up-2026). That page is the deeper dataset. This essay is how I would use it as a founder: what the 527 percent changes in the plan, what Profound's round is evidence of, and what would make me slow down. ## The demand numbers I would put on one page AI-referred web sessions across 19 GA4 properties grew 527 percent year over year, from 17,076 in January to May 2024 to 107,100 in the same months of 2025 ([Previsible's AI Traffic Report, as reported by Search Engine Land](https://searchengineland.com/ai-traffic-up-seo-rewritten-459954)). That is referral sessions in a small sample, not query volume. Beside that, [McKinsey's October 2025 paper](https://www.mckinsey.com/capabilities/growth-marketing-and-sales/our-insights/new-front-door-to-the-internet-winning-in-the-age-of-ai-search) on the new front door of the internet says half of US consumers polled intentionally seek out AI-powered search engines, with a majority of those users calling it their top digital source for buying decisions, and projects $750 billion in US revenue flowing through AI search by 2028. A February 2026 paper on global exposure reports that, in its 2025 sample of search queries, 67 percent of those run in the United States were answered by an AI Overview ([arXiv:2602.13415](https://arxiv.org/abs/2602.13415)). Google AI Overviews expanded from 7 countries to 229 in one year, on that same exposure research. I would not wait for our rankings to break before I treated this as our market. McKinsey's analysis says unprepared brands may see traditional-search traffic fall 20 to 50 percent. If the queries move, a stable rank on a shrinking commercial query is a quiet miss. The terms I would add to the scoreboard, citation and the difference between a rank and a recommendation, are in the [glossary](/glossary). The layer I would add beside SEO is the one on [Machine Relations](/machine-relations). ## What Profound's round tells me, and what it does not Profound raised $96 million at a $1 billion valuation, 18 months after founding, in a Series C announced February 24, 2026. The company tracks how AI models talk about brands. The customers named with the round include Target, Walmart, Figma, Ramp, MongoDB, U.S. Bank, and Chime, and the company said it serves more than 10 percent of the Fortune 500. The chief executive's line, reported by MarTech in March 2026, was that AI search is the biggest platform shift in the history of marketing. I would use the round as a category signal. Investors wrote a $96 million check into measurement of a behavior that Previsible's 527 percent jump in AI-referred sessions, in a 19-site sample, points to. I would not use the round as a reason to buy Profound, or any other tracker, before I can say which queries I need to see. The customer list tells me large buyers are purchasing infrastructure while the gap is still closeable. It does not tell me our citation rate. Set next to $750 billion of projected revenue and a 527 percent AI-traffic jump, the thesis is at least pointing at a real shift in demand. A valuation is not a measurement of our brand. ## The funnel I would plan for A traditional path still exists: a search for the best tool in a category, several results, several tabs, several sessions. The path I would also plan for is one question with the constraints in it, one synthesized answer, and a decision or a refinement. Discovery, evaluation, and recommendation collapse into the answer. Our presence in that answer depends on being cited by sources the model trusts, on pages that state the claim cleanly, and on showing up in the comparisons the model surfaces. Rank number one on the old page is a different problem. Gartner predicted in February 2024 that traditional search volume would drop 25 percent by 2026. This essay is being written in 2026, with that prediction on the clock. I would not pretend I have a new census of the 25 percent. I would notice where the shift lands first. The original reading of this moment says AI search takes the high-intent commercial queries earliest, and the long tail stays on traditional search longer. Overall traffic can look acceptable while conversion quality erodes. By the time the traffic line is obvious, a competitor who moved while AI-referred sessions were growing that fast, even from a small base, will have had the months of citations that make the next citation more likely. McKinsey's September 2025 survey of roughly 30 Fortune 500 consumer-brand CMOs found 16 percent systematically track AI search performance. It is a small sample, but it is the opening I would actually buy against. Most of those CMOs were not running a system. A founder who audits, records the gaps, and tracks them is ahead of the rest of that survey, even if the AI-referred traffic growth is already the market. ## The decision I would not pause SEO. I would add a parallel layer and review it on the same cadence as rankings. Scope: our brand plus our category, asked in ChatGPT, Perplexity, Claude, and Gemini; a list of trusted sources that cite competitors and do not cite us; and a rule that vague marketing copy does not count as the page we hope will be synthesized. I would track whether we are named and cited. McKinsey's 16 percent, small survey though it is, is the reason this can still look like a habit rather than a rescue. McKinsey's 20 to 50 percent traffic scenario is the reason I would not wait for the analytics to look dramatic. Profound's $96 million round is the reason the measurement category will be sold to us all year. The 527 percent AI-referral jump, small sample and all, is the reason I would believe the sellers about the direction of demand and still build our own query list. What would change my mind: a year of our commercial queries in which AI-referred and AI-answered demand stayed immaterial next to traditional search, against a market where half of consumers already seek out AI search and 67 percent of sampled US queries in 2025 were answered by AI. If that were our data, I would keep the audit and shrink the spend. A further McKinsey or equivalent read that put systematic tracking well above 16 percent would tell me the opening had narrowed, and I would compete on the quality of the sources rather than on the fact of measuring at all. I would not treat a $1 billion valuation, by itself, as a reversal or as a mandate. The expensive version of this shift is trying to start the citation habit after someone else has the sources. I would start while 16 percent is still the early-mover picture in that survey. ## FAQ ### Where does the 527 percent come from? The Previsible AI Traffic Report, which found AI-referred web sessions across 19 GA4 properties up 527 percent year over year: 107,100 in January to May 2025 against 17,076 in the same months of 2024. [Search Engine Land's write-up by Previsible's David Bell](https://searchengineland.com/ai-traffic-up-seo-rewritten-459954) is the source I read; its introduction calls it a rise between January and May 2025, but its method compares those months with the same months of 2024. It measures referral sessions in a small sample, not query volume, and [our register of seven AI-search figures](https://machinerelations.ai/research/ai-search-statistics-traced-to-primary-source-2026) records what it does and does not say. I would put it next to McKinsey's finding that half of consumers polled intentionally seek out AI-powered search and the $750 billion US revenue projection for 2028, and next to the February 2026 finding that 67 percent of sampled US queries in 2025 were answered by AI. They are different denominators. Together they say the demand moved. ### Why mention Profound's raise at all? Because a $96 million Series C at a $1 billion valuation, 18 months from founding, with customers including Target, Walmart, Figma, Ramp, MongoDB, U.S. Bank, and Chime, is a picture of who is buying measurement. On the company's account, it serves more than 10 percent of the Fortune 500. I would treat it as category formation. Our plan is still the audit and the sources, not a purchase of their software by default. ### Did Gartner's 25 percent drop already happen? Gartner predicted in February 2024 that traditional search volume would fall 25 percent by 2026. I am not publishing a new measurement of that 25 percent here. I am publishing the demand facts I can point to, and the scenario McKinsey named: unprepared brands that may lose 20 to 50 percent of traditional-search traffic, with the commercial queries moving first. ### What would I do in the next planning cycle? Audit the four engines on brand and category questions. List the sources that cite competitors and do not cite us. Replace vague pages that we hope will be synthesized with pages that state a claim. Track the result. Sixteen percent of the roughly 30 Fortune 500 consumer-brand CMOs in McKinsey's survey already do the tracking systematically. I would rather be in that group while SEO still looks healthy than explain a later traffic line with a query shift we watched from the old scoreboard. ## Correction, November 12, 2026 An earlier version of this essay said AI-referred web sessions grew 527 percent between January and May 2025. In Previsible's report, as Search Engine Land published it, the 527 percent compares the first five months of 2025 with the same months of 2024: AI-referred sessions across 19 GA4 properties went from 17,076 to 107,100. The article's introduction calls it a rise between January and May 2025, but its method is a year-over-year comparison, so the essay now says year over year. The figure and what I would do about it are unchanged. --- *Published March 26, 2026* By [Jaxon Parrott](https://jaxonparrott.com) — Founder & CEO, [AuthorityTech](https://authoritytech.io) Updated 2026-11-12 Part of [Machine Relations](https://jaxonparrott.com/machine-relations) · [Glossary](https://jaxonparrott.com/glossary) · [Founder judgment](https://jaxonparrott.com/topics/founder-judgment) Tags: AI search, Machine Relations, Founder judgment ## Continue reading - [The Category Ownership Trap](https://jaxonparrott.com/blog/category-ownership-trap): Owning a category definition does not mean owning the buyer journey. How I connect Machine Relations to the questions people ask before they know its name. - [Which Layer I Would Fix Before a Merchant Protocol](https://jaxonparrott.com/blog/agentic-merchant-protocol-product-catalog-control-ai-agents-2026): I would govern the product facts a shopping agent repeats, then audit the sources it retrieves, before I treat a merchant protocol as the visibility win. - [When AI Cannot Recognize the Business You Built](https://jaxonparrott.com/blog/why-founders-lose-ai-visibility-before-they-lose-demand): A company can have customers and demand while AI search misidentifies it. Fix conflicting identity, missing evidence and weak retrieval before buying more content. --- *Machine-readable version of [https://jaxonparrott.com/blog/the-527-percent-inflection-point](https://jaxonparrott.com/blog/the-527-percent-inflection-point)*