Seer's AI "Need-To-Knows" for June 2026
1. Bing Webmaster Tools now gives you grounding queries. It's the first to do it.
What's happening: ChatGPT, Claude, Gemini, Perplexity, Copilot: none of them give you the prompts, and none gave you the grounding queries either. Bing is the first one. Google added AI reporting two days later, but it's likely "mostly useless" because it isn't giving you queries.
What to do about it: Set up Bing Webmaster Tools and read your grounding queries. When one surfaces an old page, make the page better. We've got to turn that visibility into believability, not just another citation in a report.
2. Personalization is now a search input.
What's happening: Garrett Sussman found that once you connect personal intelligence to your Google account, brands got mentioned more in his AI answers when he had receipts in Gmail or a brand of hoodie in his Photos. Google's had Gmail, Photos, and Calendar for years and never used them. Now, with competition from ChatGPT and Claude, they are.
What to do about it: Start treating every search as an MSV of one. Ask how you're partnering with the team sending your emails, or deciding where your logo gets imprinted. Those are inputs now.
3. Reviews are a huge part of AI discoverability, and most of us don't own them.
What's happening: In a Trustpilot-commissioned study, even a thin profile beat no profile by 52 percentage points in AI search visibility.
What to do about it: This isn't a "go buy Trustpilot" recommendation. Audit your third-party review presence and fix your zero or near-zero profiles. You have to lead a horse to water here: meet customers where they are and build a real mechanism to solicit reviews.
4. Stop chasing citations as a success metric. They're chaotic, and Reddit is surging.
What's happening: What's visible on ChatGPT is very different from Perplexity, Claude, and Google's AI Overviews, and it shifts with every new model. Matt Boxbom found a ridiculous surge in Reddit's influence on Google's AI Overviews specifically.
What to do about it: Don't go off on an island. Correlate citation data to what your customer actually engages with. If that's Reddit, stand up a real, resourced presence.
5. AI Overviews and AI Mode are two different behaviors. Track them separately.
What's happening: Kevin Indig's research shows AI Mode users mostly take the first answer (around 74% go with whatever's listed first), while AI Overview users behave more like browsing Netflix: scrolling down, then back up, weighing every citation. Google now reports AI Mode at ~1 billion weekly active users.
What to do about it: Split AIO and AI Mode into two optimization tracks with their own metrics. Treat AIO presence more like a CPM impression than a click, and apply a meta-description mentality to your titles so your UVP lands without a visit.
6. Amazon's Alexa for Shopping: discovery and purchase are moving inside the agent.
What's happening: Amazon made agentic commerce mainstream overnight, to an audience where few even know what an agent is. It's not a search portal, it's an action portal: find me the best deal, remind me when I'm low.
What to do about it: Pressure test your category on Alexa for Shopping, even if you don't sell on Amazon. See who's winning.
7. Google I/O: the minimalist roots are long gone.
What's happening: Google's biggest search-bar change in 25 years: a choose-your-own-adventure experience with Gemini powering the bar, and AI Mode now at ~1 billion weekly active users.
What to do about it: Educate your execs, because an old mental model of Google leads to wrong investments. Getting added as a preferred source could be a minor cheat code.
The Q&A Doesn't Stop When the Webinar Does
Q: Best ways to increase visibility and authority in Google's AI Mode specifically, versus traditional SEO? Do we even know? A: Wil doesn't go deep on entities or EEAT, because once you name it, isn't it just about trusting the source? That's just marketing. What he sees in his weekly "tell me about Seer Interactive" checks is that AI looks for differentiators (B Corp is probably an entity). Try mapping the grounding queries Bing gives you to entities, then test whether saying those things more changes your answers across all AI, not just Google. Alisa: don't chase the data to the nth degree. By the time you crack the case, a new model's out or your audience has shifted.
Q: Tips to track AI Mode? We're tracking AI Overviews already. A: It's all about buying data right now. If you already track GEO, you can probably buy it from a vendor like DataForSEO affordably. Beat people up on price and don't take the first offer. Even a benchmark matters for looking back in 6–12 months. Wil's frame (via Seer Alum Adam Nelson): your job is to shrink the black box. Even running a prompt 10 times shrinks it enough to give your manager an answer. One basic catch (via Jason Thompson): ChatGPT's mobile app sends different referral data, so a mobile audience can make your numbers look smaller than they are.
Q: Won't Microsoft share grounding-query data with OpenAI given their relationship? A: Could be. Bing says "Copilot and partners," and OpenAI is their biggest partner, but they aren't saying it explicitly. Maybe we test that.
Q: If you have nothing personal to share, is it better to publish nothing than AI-generated content? A: Take that all the way up the org and ask: what company have we built where everything could be copy-pasted onto a competitor's site? Most teams do have points of view, they just never built a pipeline from internal to external. Shout out to our alum Greta, who mines her internal Slack with Claude's connector and gets content 30–40% done before she starts.
Q: Was that an HTML page that looked like markdown, and is it in your main nav? A: It's an HTML page (/ai-information), not a text file. It's linked only in the XML sitemap, which has Wil thinking internal linking probably matters less than we think in an AI and agentic world.
Q: Any correlation between writing style/tone and AI visibility? AP-style versus casual? A: Wil's a good test: his writing is full of rap references, swear words, and typos, and he doesn't see it hurting his visibility. He'd rather connect in a way that makes people feel the webinar's worth showing up for. Spicy language from passion isn't always professional, but nobody ever showed up to a speaker because they used AP style.
Q: We have reviews on third-party sites. Worth collecting the best ones on our own page, like Seer's testimonials page? A: No reason not to. It's worth a test even if it does nothing for 3–6 months. It's a huge pain to get reviews onto third-party sites, so controlling more of that influence with something you own is all the more reason to do it.
Q: Does the primary review source matter by industry? Trustpilot isn't big in travel. A: Yes. Look at where your citation data comes from. There's an invisible criteria baked into the model's reasoning that picks the best source (TripAdvisor, Google reviews for travel). Wil's add (via Chris Long): ChatGPT runs site-colon searches against trusted domains (G2, Gartner, Trustpilot in B2B) to filter spam. Find the query fan-outs, see which domains it restricts to, and treat those as your priorities.
Q: Seed Reddit conversations, set up a company subreddit, or just join existing ones? A: Treat it as a maturity model, not 0 to 100. Tread lightly (your CMO may think it's a quagmire). Start with smaller subreddit communities, clarify objectively false statements, be genuinely helpful. The goal isn't to answer "best GEO agency?" with "it's Seer." It's to add value, same as any marketing. More coming via the newsletter.
Wil and Alisa's Latest Thoughts
Seer Interactive Launches Local GEO Academy
Yes, we're paying young people to learn and test improving AI results A lot of companies right now are cutting entry-level roles. Some are replacing them with A…
Finance Industry Giants Can No Longer Rely on Authority: Here's How Clarity of Content Is Becoming the New Competitive Edge
In a recent usability study our UX team completed with Outset.ai we found that users are most comfortable with tasks throughout the research and consideration p…