Google Just Let Your Audience Vote on Your AI Visibility

On August 20, Google shipped something no AI search surface has had before: a button you can embed on your own website that lets a reader tell Google to rank you higher. One click, and that reader sees more of your content in Top Stories, in AI Overviews, and in AI Mode, labeled as coming from a source they chose (PYMNTS).

Sit with that for a second. Every AI visibility tactic of the last two years has been aimed at machines. Structure your content for retrieval. Mark it up with schema. Get mentioned on the sites the models trust. All of it is supply-side work: you reshape what the machine reads and hope the answer changes. The Preferred Sources button is the first demand-side lever. It doesn’t persuade the model. It lets your audience overrule it.

Most brands will ignore this because it’s framed as a publisher feature, something for newsrooms fighting traffic decline. That framing undersells it. What Google actually built is a mechanism that converts audience loyalty directly into AI answer placement. If you have a newsletter list, a returning readership, or a community that actually likes your content, you now have a ranking input your competitors can’t copy by hiring a better agency.

What actually shipped

Preferred Sources itself isn’t new. What’s new is where it reaches and who controls the prompt to use it.

Date	What happened	Sources selected
August 2025	Preferred Sources launches for Top Stories	—
December 2025	Global rollout to English-language users	~90,000
May 2026	Extended into AI Overviews and AI Mode, with visible “Preferred Source” labels	~345,000
August 2026	Embeddable on-site button ships	600,000+

The May expansion was the quiet turning point: from that release on, marking a site as preferred started changing Google’s AI answers themselves, beyond the news carousel (Search Engine Journal). But users had to dig through Google’s own settings to select anyone, which is why adoption sat in the hundreds of thousands, not millions.

The August change removes that friction. The button lives on your page, at the moment of maximum goodwill: right after someone finishes an article they came looking for. The reader clicks, confirms, and lands back where they were. No detour through Google settings. Adoption numbers already reflect it: Mrinalini Loew, who runs Google’s search ecosystem team, put the count at more than 600,000 unique sources, up from roughly 345,000 in May.

And Google is telling you the payoff outright: people are twice as likely to click through to a preferred source when one appears. Hold on to your skepticism about that number (we’ll come back to it), but notice what Google is signaling by publishing it. They want this used.

Why this lever is different

Think about what every other item on your GEO checklist has in common. Content structure, entity clarity, schema, digital PR, Reddit presence, Wikipedia — each one is an argument you make to a retrieval system, and the retrieval system decides. You can do everything right and still lose the citation to a competitor, because the decision was never yours.

This signal moves the decision to a third party you actually have a relationship with: your reader. Loyalty stops being a fuzzy brand-marketing outcome and becomes retrieval bias, per user, compounding as your audience grows. A competitor can match your schema in a sprint. They cannot match ten years of newsletter subscribers pressing a button.

It also arrives at a convenient moment for Google. Search referral traffic keeps sliding as the company pushes users from AI Overviews toward the fully conversational AI Mode, and publishers keep saying so loudly. Handing sites a visibility button is a cheap way to answer that criticism: if your traffic fell, well, did you ask your readers to prefer you? Cynical or not, the deflection works in your favor. Google needs this feature to visibly succeed, which means it’s likely to get more surface area, not less.

The fine print Google isn’t printing

Three caveats before you get excited.

First, the 2x click-through stat is selection-biased. People click sources they chose because they chose them. The stat measures loyalty being expressed, not loyalty being created. Adding the button doesn’t manufacture preference; it harvests preference you’ve already earned. If nobody would press it, that’s a brand problem the button can’t fix.

Second, it’s a rich-get-richer mechanism. The brands best positioned to benefit are the ones with the largest existing audiences. A challenger brand with 400 newsletter subscribers gets a rounding error; a national publication gets a moat. If you’re the smaller player, this lever matters most on your home turf: the niche audience that genuinely prefers you over the incumbent. Concentrate the ask there.

Third, it was designed for publications, and Google hasn’t said how far it stretches. The rollout language is all “publishers” and “newsrooms.” Nothing in the mechanics restricts it to news: readers can add any site they choose, and Google’s own guidance says sites can prompt their readers to do it. But how much weight a preferred B2B blog carries in a commercial AI Mode answer is an open question nobody outside Google can answer yet. The cost of finding out is an afternoon of implementation. Take the trade.

Your dashboard can’t see any of this

Here’s the part that should bother you, and it’s the reason this story belongs on a monitoring company’s blog.

Preferred Sources is a personalization signal. It changes what one specific user sees. Which means your AI visibility just split into two different realities: the boosted answers your loyal audience gets, and the neutral default everyone else gets. No report Google gives you distinguishes them. Search Console’s AI report already shows impressions without clicks or queries; it certainly won’t tell you which impressions came from preference boosts. This is the same measurement problem we flagged when Gmail and Photos started feeding AI Mode answers. Personalization keeps making “our brand’s visibility” a less coherent single number.

Most teams are in no shape to handle that split. Semrush’s 2026 AI Visibility Index, built on 126 million U.S. prompts from January through April, found that 45% of marketing leaders can’t accurately measure brand visibility in AI answers, and only 9% have tooling that covers the metrics they need (Semrush).

The way through is to be precise about what each measurement represents. Monitoring tools query AI platforms from a neutral, logged-out baseline. That baseline is exactly what your preference boost doesn’t touch, and that’s a feature, not a gap. The neutral default is what new buyers see, the people who haven’t heard of you and would never press your button. Preferred Sources can’t win them; it only deepens your hold on people already won. So the two views answer different questions. The boost retains; the baseline acquires.

If your neutral-baseline visibility is weak, a successful button campaign will hide the problem from your own analytics while every net-new buyer still gets an answer without you in it. And since a single prompt was never enough to measure the baseline anyway, you need systematic tracking on one side of the ledger precisely because you’re now pulling a personalization lever on the other.

What to do with it this week

Embed the button. Google published implementation instructions with the launch. This is an afternoon of work, most of it deciding placement.
Put it where loyalty peaks. End of articles, post-signup confirmation pages, the thank-you screen after a purchase or download. Not your homepage: visitors there haven’t earned the ask yet, and you haven’t earned the click.
Run it through your owned channels. One dedicated newsletter mention explaining what readers get (your content, labeled and prioritized, in their Google results) will outperform any passive placement. Your list is the whole game here.
Snapshot before you start. Record your Search Console AI Mode impressions and your neutral-baseline visibility now, so you can attribute any movement honestly instead of guessing in three months.
Keep the baseline honest. Track your brand’s default AI answers on a schedule, across prompts and platforms. That number is your acquisition reality, and the button will never move it.

The signal behind the signal

The interesting thing about Preferred Sources isn’t the button. It’s what Google is admitting by building it: the model’s own judgment about which sources to trust isn’t good enough to defend, so Google is letting users patch it, one preference at a time. Human trust is being wired into AI answers as an explicit input because the implicit version kept getting the blame.

Expect this pattern to spread. ChatGPT’s memory already nudges its answers toward brands you’ve engaged with, just without an explicit vote. Once one platform makes preference a promotable signal, the others have every reason to follow. An audience that can vote for you is an audience protected by switching costs. The brands that win that world aren’t the ones with the best schema. They’re the ones whose readers, when handed a ballot, bother to fill it in.

RivalHound tracks your brand’s visibility across ChatGPT, Google AI, Perplexity, and more. Start monitoring to see where you stand.