The page ranked first. Had ranked first for eight months, through two algorithm updates, with a solid backlink profile and the kind of structured content that every technical SEO checklist would approve. But the AI Overview for that exact keyword was citing a three-year-old article on a domain with a fraction of the authority, no recent updates, and markup that looked like it had been copied from a template. My client asked me to explain it. I told him I'd get back to him, which is the honest version of saying I didn't have a clean answer yet.
I had one now, sort of. The antitrust court filings from the DOJ case against Google have been quietly producing disclosure documents that the SEO industry is only beginning to parse properly. One of the more significant ones involves a system referred to internally at Google as "FastSearch." The broad picture that emerges from the filings is that Google does not use a single monolithic index to generate AI Overviews. It uses a separate, faster retrieval system, built for speed rather than comprehensive quality evaluation, to ground the responses its AI generates before serving them to users.
This matters more than most of the coverage has made it sound.
The main Google index, the one that drives traditional organic rankings, is built around quality signals that have been refined over decades: PageRank derivatives, link authority, topical expertise signals, E-E-A-T frameworks, user engagement data fed through systems like Navboost. The FastSearch index, as described in the filings, is optimized for retrieval speed. It is smaller, structured differently, and does not apply the same depth of quality evaluation. Which means the signals that make a page rank well in organic search are not necessarily the signals that get a page cited in an AI Overview.
Image credit: Screenshot from "New Report: Google AI Overviews Cause a 60% Traffic Drop for Small Websites!" by BrenTech on YouTube (https://www.youtube.com/watch?v=ujTowFCnGVw).
That is not a subtle distinction. That is the foundation of two entirely different optimization problems being sold to clients as if they are the same problem.
The conventional advice right now is to optimize for AI Overviews by improving your E-E-A-T signals, strengthening your topical authority, earning better links. That advice is correct for traditional organic rankings. For the FastSearch index, the signals that appear to matter more are structural: how cleanly the page can be parsed at retrieval speed, how directly the content answers a specific query at the sentence level, whether the markup is tight enough for rapid extraction without disambiguation.
Here is where I got it wrong before the filings confirmed any of this. I had a SaaS client with genuinely strong content. Long-form, well-researched, properly linked. We spent four months improving their topical authority cluster. Traditional rankings improved. AI Overview citations for their target queries did not improve at the same rate, and in two cases got worse as other, thinner pages from faster-loading domains started appearing instead. I assumed we had a freshness problem and started pushing more frequent content updates. That helped slightly. But the real variable, which I only understood in retrospect, was structural parsability. The pages getting cited had one thing in common: the answer to the query appeared in the first two sentences of a clearly labeled section with matching header text. Not because of any schema trick. Because the FastSearch retrieval pass could find it without working.
The SEO agencies currently selling "AI Overview optimization" packages are, almost uniformly, optimizing for the wrong index. The reports they send clients show improvements in traditional authority metrics and use those improvements as evidence of AI visibility progress. Those metrics are not measuring the same system. You can have a hundred referring domains and a perfect Core Web Vitals score and still not appear in a single AI Overview if your page structure cannot survive a high-speed retrieval pass.
This is also why the publisher blockade conversations have been so noisy without producing much useful guidance. Publishers blocking their content from the FastSearch index are opting out of a different system than the one their SEO reports are measuring. The authority signals they spent years building for traditional search barely transfer.
The most important thing the FastSearch filings revealed is not a scandal. It is a calibration problem. There are now two optimization targets sitting inside what most people call "Google SEO," and they are being measured, priced, and sold as if they are one.
Whatever work you do for traditional rankings, treat AI Overview visibility as a separate audit with separate criteria: answer structure, retrieval parsability, query-to-sentence alignment in the first visible paragraph. If your current SEO provider is not running those two audits separately, they are billing you for one job while telling you they are doing both.
Waleed Qamar holds a BSc in Computer Science from Purdue University and has spent the years since turning that technical foundation into something the curriculum never covered: figuring out why websites rank, why they fall, and why most businesses never find out until it is too late.
Pakistan-born and based between the United States and South Asia, he has managed search visibility for e-commerce stores, local service businesses, and SaaS startups across two continents. He started in SEO when guest posting still worked, survived the Penguin update, and has rebuilt client sites from scratch after algorithm hits more than once.
He has watched good businesses get sold packages that looked like progress and delivered nothing lasting. He has also seen the right approach quietly double a site’s traffic without a single press release about it.
His writing on SEO By Highsoftware99 covers Google algorithm updates, autocomplete optimization, semantic SEO structure, and the widening gap between what agencies promise and what Google actually rewards in 2026.
He knows what a traffic cliff looks like in Search Console on the morning you discover it.