Microsoft Clarity's New AI Citation Filter, LinkedIn's B2B Guide, and Claude Has a New Watermark

📬 You like this podcast? You’ll love the newsletter.
Join the weekly The Visibility Report: subscribe

A few stories with real strategic implications, and one that's generating more heat than light this week.

First: Microsoft Clarity adds branded vs. non-branded query segmentation to its AI Citations dashboard. Cassie breaks down what the new filtering actually shows you, why separating brand-led demand from category discovery changes how you interpret your citation data, and what Share of Authority looks like when you can finally split it by query type.

Then: LinkedIn publishes its first comprehensive B2B AI search guide, backed by internal data and eighteen months of their own testing. Cassie pulls out what's actually useful, including why LinkedIn is the number one most-cited domain for professional queries in AI search, how articles and posts do different jobs in the citation ecosystem, and the org alignment point that most brands are still missing.

Plus: Sundar Pichai announced Gemini hit one billion monthly users, Shopify's Q2 data shows AI-referred sessions growing 197% year-over-year with double the conversion rate in research-heavy categories, and the Claude watermarking story — what we actually know, what's still speculation, and why you shouldn't change your workflow yet.

In this episode:

• What branded vs. non-branded grounding query segmentation actually tells you
• How to use Share of Authority data now that you can filter by query type
• Why LinkedIn articles and posts do different jobs in AI search
• The org alignment problem LinkedIn's own guide surfaces
• What Profound's citation timing data means for how you evaluate new content
• What Gemini's milestone signals about where AI search is headed
• What Shopify's Q2 data says about buyer behavior in research-heavy categories
• What we actually know about Claude's watermarks — and what's still noise

I'm an AI search visibility consultant and a fractional content strategist for startups and enterprise brands. If that sounds like the kind of help you're looking for, email me at cassie@cassieclarkmarketing.com. 

Or request your AI Search Visibility Audit: https://cassieclarkmarketing.com/ai-search-visibility-audit/

Let’s connect:

LinkedIn → Cassie Clark | AI Search Visibility Consultant
Website → https://cassieclarkmarketing.com

Transcript

SPEAKER_00 0:31

Hey, welcome back to Found in AI. I am Cassie Clark, an AI Search Visibility Consultant and host of the show where we track what's actually happening in AI Search and what it means for your strategy. Today is Thursday, August 13th. We have a few updates to chat about this week. Let's just get right into it. First up, Microsoft Clarity. They just added something helpful to the AI citations dashboard last week. When you look at your AI citation data, not all of those citations are coming from the same kind of query. Some of them are happening because someone asks an AI system something that directly referenced your brand. Something like, what does Cassie Clark marketing do? You know, that kind of thing. Others are happening because the AI system pulled your content to help answer a broader generic question. Sometimes those questions don't really have anything to do with your brand, they're not really direct, but it's something that your brand talks about. Those are two completely different things. However, without a way to separate them, you're looking at a blended number that doesn't really tell you much. That's what this update fixes. The dashboard now distinguishes between branded and non-branded grounded queries. You can see which individual queries are brand specific, and there's an AI share of authority breakdown by query type, and you can filter the whole dashboard to compare how you perform on branded versus non-branded queries. So a question I like to ask is why does this matter? Well, because the two signals diagnose different problems. Strong branded query citation tells you about direct brand awareness. People already know that you exist and they are asking about you specifically. Strong non-branded citations tells you about category authority. Those AI systems are pulling you in as a source even when your brand isn't a part of the question. That's the one that's harder to earn, but it's gonna be the one that lasts a little longer once you finally earn it. If you've been tracking your AI citations and being webmaster tools or clarity and then feeling like the numbers aren't really telling the whole story, I think we've all kind of felt that way with AI citations in general. The update is gonna help fix some of that. Alright, next thing. LinkedIn published a guide this week called Unlocking AI Search Visibility, the B2B Marketer's Guide to LinkedIn. It's authored by their VP of marketing, Devink Shaw. This is heftier than just post more on LinkedIn. There are a few things that stood out though, and we need to talk about them. The first is a stat from Profound's 2026 research confirming that LinkedIn is the number one most cited domain for professional queries in AI search. That's kind of a big deal. If you're a B2B brand and you're not treating LinkedIn as an entity building surface, not just a distribution channel, this is something worth thinking about. The second is the breakdown of how articles and posts do different jobs. According to LinkedIn's own data, articles generate roughly 60% of LinkedIn content citations, and posts account for the other 40%. Articles are your death play, they build the long-form authority AI models draw from. Posts are your distribution play, they create those signals and surface the themes that you want to be associated with. You need both of them and they feed each other. The third thing that I want to pull out is something from their head of organic growth, which I think is really worth repeating here. She said don't fall for universal best practices. What works for one brand might not work for yours. I think that is true for anything, and I say this quite often. The framework gives you something to start with, but your own testing tells you what's actually working, and trust your own data. There's one other data point that's mentioned in this guide that should probably get more attention than I'm sure it will. Profound found a median of 6.81 days for new pages to be cited by Chat, GPT, or Claude, but 90% of pages take up to 37 days. So if you publish something and then check your citation data a week later and see nothing, I am really guilty about this. It's not a signal that it didn't work. You just need to give it more time. 30 days is usually about the sweet spot. The guide also validates something that I have been saying for a long time now. AI visibility is not a content team problem. I will say it until I'm blue in the face. And if you're tired of hearing it, I'm sorry you're gonna hear it a hundred times more. LinkedIn built an internal AI search task force that pulled in SEO, PR, editorial, social, paid, brand, and product marketing. That is the org alignment piece that I'm talking about. One team cannot own this. The signals come from absolutely everywhere. Alright, moving on. Let's talk about scale for a second. Sundar P Chai, he is the Google guy, announced this week that the Gemini app has hit 1 billion monthly active users. This is Google's fastest growing product ever, and it's their 14th product to reach that milestone. On its own, that's a big number, but I want to pair it with what Shopify just published from their Q2 Commerce data because when you hear those numbers together and then you hear the Shopify information, you're like, oh, oh, that's a bit of a light bulb moment. Shopify tracked AR referred sessions to merchants on their platform and found they grew 197% year over year in Q2. Organic Search grew just 12% on a much larger base. Both of those are growing, so this is not a zero-sum dynamic that we are dealing with here. The Search Pi is just getting bigger and it's spreading across more surfaces. But here's what's interesting about where AI Search is performing best. In research-intensive spec-led categories, so like places where buyers need to compare options and check compatibility and evaluate trade-offs before they make some kind of purchasing decision. AI referred shoppers converted at roughly twice the rate of organic search visitors. Shopify's interpretation is that AI is compressing the buyer journey. Someone uses AI to do the research, they arrive at the product page already pretty much decided, and then they convert. I talked about this a bit yesterday in the LinkedIn newsletter, and if we're not connected, I would love to be, I'm just saying. But back to that Shopify data. When AI drew from structured catalog data, the conversion rate was two times better than when it relied on scrapped or third-party fiends. So the quality and structure of your underlying content isn't just a visibility question. You also need to think about conversion too, which sounds silly, like we're marketers, we think about it, but we can't forget that part. This also maps directly to B2B. High consideration purchases, whether that's software, services, consulting, enterprise vendors, whatever, that involves exactly the kind of research heavy evaluation where AI is doing the work earlier in the funnel. So by the time someone reaches you, they've already formed expectations. Your content needs to hold up to that scrutiny, not just earn the click. Like it really needs to shine once I get there. The Gemini milestone is pretty much just telling us that the surface really is growing and we should really start to pay attention to it. The Shopify data tells us that the behavior on that surface is also maturing. When we mix them together, this is not a wait and see moment. Like this is something that we need to really start working on now. And then there's Claude. I cannot wrap up a week in AI search without mentioning something that has been all over the internet this week. Anthropic announced that Claude will be adding watermarks to AI generated text. The short version of all of this, like the TLDR version, is that new Claude models will embed an imperceptible watermark directly into the text that they generate. You won't see it. It doesn't change how the content reads, but it does travel with the text. And here's the part that some people are really calling out and then other people are just kind of glossing over. It applies not just to text that Claude fully generates, but to any text that Claude processes. So if you paste something in for a proofread, something that you wrote yourself and you need a second set of eyes on it, and then Claude corrects a few typos, that output can carry a watermark too. There is a ton of speculation right now about what all of this means for content creators, for SEOs, for AI search visibility. Everybody and their brother has an opinion on it. I mean it's absolutely everywhere. And I want to be kind of just honest with you here for a second. The guidance about all of this just isn't there yet. Anthropic mentions that they have a tool coming out that users in third-party platforms can use to check for this AI-generated watermark, but at the moment no one really knows how this will shake out. Does this become a ranking signal? Who knows? Does it factor into how AI engines evaluate trust and providence? Maybe. Maybe, eventually, who knows? But right now we're in the wait and see period. And I can't really tell you what the implications are gonna be because, like I said, we just don't know right now. But what I will say, and I say this because I know a lot of marketers are using Claude or Claude Code or Cowork or whatever in their workflows, I don't think I would change that up yet just based on speculation. Just watch the space and see what happens. Things need to shake out first before we go changing up things again. Now, before I let you go, if you want to go deeper on any of this, the visibility report, that's my Friday newsletter goes out well every Friday. It's where I do a longer analysis, so it's less news and more of hey, here's what this actually means for your strategy. And starting to bring in guest writers too, so you don't want to miss that. If you're not subscribed yet, the link is in the show notes. Okay, that is it for today. I will see you on Tuesday. Until then, stay visible.