News · August 12, 2026
Google's AI Overviews gutted publisher traffic. Here's the solo-operator playbook.
The numbers are no longer arguable. Google search traffic to publishers fell roughly a third globally in the past year, and 38 percent for US publishers. Organic click-through on queries showing an AI Overview cratered from 1.76 percent to as low as 0.61 percent before partially recovering this spring. Surveyed publishers expect another 43 percent decline over the next three years. If part of your business depends on people clicking from Google to a website, this is the environment now, and pretending otherwise is a strategy for slowly going broke.
But the aggregate numbers hide the part that matters for anyone small: the damage is wildly uneven, and the unevenness is the playbook.
Where the damage concentrates
AI Overviews trigger on roughly 36 percent of informational queries. “What is human design,” “how does email deliverability work,” “what are AI agents.” Those pages get summarized, the reader gets the gist, and the click never happens. If your content strategy is explaining things, Google is now your competitor, using your own pages as raw material.
Commercial queries trigger Overviews only about 8 percent of the time, and transactional queries about 5 percent. “Best AI voice generator,” “GoHighLevel vs Mailchimp,” “ElevenLabs discount.” Google leaves those alone because summarizing a purchase decision poorly is a liability, and because those queries feed the ads business. The traffic that converts to revenue was always commercial traffic, and commercial traffic is the slice that survives.
And one more asymmetry that most coverage misses: pages that get cited inside an Overview earn roughly double the clicks per impression of uncited competitors on the same query. The Overview is not just eating traffic, it is redistributing it toward the handful of pages it quotes.
The five moves that follow from the numbers
Shift the portfolio toward commercial intent. Not entirely, informational content still builds topical authority and feeds internal links. But the pages you invest real hours in, the testing, the screenshots, the updating, should be the comparison, review, and best-for pages that Overviews mostly leave alone. On this site, that logic is why the hubs and the directory exist.
Write to be cited, not just to rank. Citation goes to pages that make clean, quotable, specific claims early. A definitive answer in the first 80 words. Stated numbers instead of hedges. One-sentence verdicts a machine can lift. It feels strange writing for a summarizer, but the citation traffic is real and doubling your clicks per impression is worth a structural habit.
Make the click irreplaceable. A summary can relay facts; it cannot relay your screenshots, your test results, the mistake that cost you months, or a verdict with your name on it. Every page should contain something a summary cannot carry. If a page has nothing a summary cannot carry, that page was already a commodity, and the Overview just made it official.
Move the center of gravity to owned channels. This is the big one. Search traffic is rented, and the rent just went up by a third. An email list cannot be algorithmed away. Every page on a modern content site should have one job beyond its topic: convert the visit into a subscription, because the visit may not repeat. Well-run niche newsletters open at 35 to 45 percent; compare that to the 2 percent CTR on an Overview query and the strategy writes itself.
Diversify discovery beyond blue links. Google Discover still sends meaningful traffic to sites with strong images, fresh publishing, and entity-rich writing. Pinterest still compounds for visual niches. Neither is a search replacement; both are hedges against a single point of failure that just demonstrated its failure mode.
What we are doing about it, concretely
This site is the playbook applied: commercial hubs carry the revenue, every page captures email for The Drop, news posts like this one are structured to be citable, and the directory targets the comparison queries that survive. None of that is clever; it is just reading the same numbers everyone can read and building accordingly. The operators who get hurt worst in the next three years will be the ones still running the 2019 playbook of informational volume, because that playbook now produces free training data for the machine that replaced it.