Goodie AI @Goodie_AI

Goodie AI now tracks Muse and Instinct. If you’re already watching what buyers ask AI, this adds the other half: what happens when an agent acts on their behalf. Now you can see exactly where your brand stands when Muse and Instinct are the ones deciding.

What happens when Siri AI can recognize you by your face? @Apple just unveiled HomeView, a smart display built around Siri AI. It recognizes up to six people in a household and switches to each person’s own Home Screen, widgets, messages, and calendar. That changes what an answer is built from. Ask HomeView about tonight’s plans, and Siri responds with the user’s own messages and calendar, so two people on the same screen get two different evenings back. For now, that context stays personal. But if AI search keeps getting to know the person behind each prompt, “does AI cite us?” won’t be the only question brands are asking. The next one might be, “Does it cite us correctly for the specific person that's asking?”

John Ternus @johnternus

2 weeks ago

Welcome HomeView

3K 6K 89K 9.0M 7K

What is your product best at, according to ChatGPT? OpenAI is testing labels on ChatGPT’s shopping results. One product can earn a green “best all-rounder” tag, while others on the shortlist get gray tags like “best for beginners.” Shoppers will likely gravitate toward the label that matches their own situation, because it feels like a personal recommendation. It answers the question they came in with: which one is right for me? The label can only be as specific as what the web already says about a product. Brands known for one clear strength give ChatGPT something concrete to point to. Do you know which label your closest competitor is earning? #AEO #AIsearch #AgenticCommerce

Curious where TikTok ranks against the rest of the field? Dive into it here: higoodie.com/blog/social-me…

Nobody builds an AI search strategy around TikTok. But our citation data may convince you otherwise. TikTok's citations grew from 170 in January to 1,641 in August, even as total social citation volume across the panel declined. That growth is spread across the board. No single AI surface holds more than 22% of TikTok's citations, the most distributed footprint of any platform we track. Reddit tells the opposite story: at points this year, a single engine accounted for as much as 90% of its citation weight. That's what makes TikTok a platform where presence pays off. Concentrated growth is a bet on one model's favor. Distributed growth compounds no matter which AI surface you're optimizing for.

Expert reviews are beating customer reviews in AI search (by a lot). We pulled over 900k citations across 31 industries to see which review sources AI models cite when someone asks what to buy. App stores and marketplaces barely register. Apple's App Store peaks at 0.42%. Google Play tops out at 0.37%. The sources actually earning citations are expert testing outlets like RTINGS and Consumer Reports, which hold 4.14% of all citations within 21 consumer industries. That’s 15 times the share held by customer-review platforms. This is Volume 1 of our Review Site Citation Study. Inside: the industry-by-industry breakdown, which review sites win each category, and where the pattern flips entirely. Read the full breakdown: higoodie.com/blog/expert-vs…

The platform an AI model trusts most today may not be the one it'll trust next quarter. ChatGPT cites Reddit 89.5% of the time. Reddit is licensed to OpenAI. Gemini leans on YouTube, which Google owns. Copilot and Meta AI both default to LinkedIn, Microsoft's platform. Claude and Perplexity, both named in Reddit's lawsuits, barely touch it. Every concentration on that list traces back to a deal, a lawsuit, or an ownership stake, and none of those stay fixed. The map reshuffles every time one of those changes. YouTube already proved it. Google supplied 82.5% of YouTube's AI citations a year ago. Today it's 42%, with Grok now the single largest source. A platform strategy built around one model's current preferences expires the moment that model's deals do. Monitor weekly, not quarterly. Download Volume 3 of our Social Study for more findings: higoodie.com/blog/social-me…

Does where you work change how you search? Yesterday, OpenAI launched Space, a shared ChatGPT workspace for teams and AI agents, plus Pages, a doc where research and writing happen in one place. OpenAI, it seems, would like what the office-suite giants have had for years. Google and Microsoft's AI advantage has always been the ecosystem: work already lives in their tools. So for anything from docs to slide decks, Gemini and Copilot have been the default AI assistants. ChatGPT had the users, but using it meant leaving your work in a separate tab. With Space and Pages, the work comes to ChatGPT, and so do the searches people run mid-task. What are your thoughts on ChatGPT's Space? @OpenAI #AEO #AIsearch #ChatGPT

ChatGPT cut Reddit citations by 66% in two weeks. That's the story every outlet covered. But it's not the full one. What they missed: ChatGPT's entire social layer contracted, not just Reddit. Social citations dropped from 6.8% to 2.5% of ChatGPT's total mix. Reddit took the visible hit because it made up 90% of that mix, the most concentrated single-platform dependency of any model we track. A similar collapse happened in September 2025 and rebounded within weeks. So no, "Reddit isn't dead." Any source making up 90% of a model's social mix is a single point of failure, and it can move without warning. If your visibility strategy leans on one platform, an engine's next update decides your reach, not you. More on ChatGPT's cut from Reddit: higoodie.com/blog/social-me…

Claude didn't cite Reddit once in eight months. Reddit is the top social source for both ChatGPT and Grok, but Claude skipped it entirely. TikTok makes up 44.9% of Claude's social citations and Substack makes up another 7.7%. However, social citations make up just 0.3% of Claude's total citations, the lowest share of any model we track. Reddit’s litigation against Anthropic explains part of the gap. Brands courting Claude's audience need a different play: have a strong named-person presence on LinkedIn, an active newsletter, and a clear TikTok profile. Reddit isn't where this model is looking. Full findings from our AI's Social Diet Report: higoodie.com/blog/social-me…

Rank tracking gave marketers a clear answer every morning for twenty years. AI search took that away. AI models don't hedge when they don't know something about a brand. They answer anyway, with the same confidence as a fact they got right. Platform feedback buttons exist, but corrections aren't guaranteed and don't happen fast enough to matter before a prospect decides. That's the gap Brand Command closes. Define the facts about your brand once. Every AI answer gets checked against them across models. Brand Command is available now → higoodie.com/features/ai-br…

keyboard_arrow_left Previous keyboard_arrow_right Next

AI share of voice is often treated as a single metric. It's two, and most brands only watch one. AI share of voice measures how often a brand shows up in AI answers. It splits into two signals, depending on what you count. Mention share of voice tracks how often a brand's name shows up. Citation share of voice tracks how often its content gets directly attributed as a source. Tracking mention rate alone gives a false read on where a brand stands. Goodie's Visibility Monitoring tracks both mentions and citations across every major AI model. Check it out: higoodie.com/features/ai-vi…

keyboard_arrow_left Previous keyboard_arrow_right Next

Volume 3 of our social citation study just dropped: AI’s Social Diet. We tracked eight months of citations this time, our longest window yet, to measure something the first two studies could only hint at: how fast the social citation landscape shifts. Download the full study here: lnkd.in/eqNFqDVm

This week, Gemini told hikers to pack far less food and water than they needed for their climb up Mount Shasta. That's how you turn an 8-hour hike into an overnight rescue mission. The citations AI models collect directly affect someone's decision-making. Your brand is defined by what gets said about you across the internet. AI just repeats it back with full confidence. Get that story wrong, and someone else pays for it.

TechCrunch @TechCrunch

2 months ago

The sheriff’s office said the hikers “were advised by Gemini to bring far less food and water than their group required." spr.ly/6013BGDp8R

21 18 83 84K 17

Most AEO checklists tell you where to click. They don't tell you what any of it's worth. Our data shows owned product pages account for 1.7% of what AI models cite when they recommend a product. The layers that carry the real weight are the ones most roadmaps treat as secondary: entity consistency and earned coverage. A brand can perfect its own product pages and still lose the shortlist to a competitor who invested in the other 98%. Measurement exposes the mismatch. Without it, a brand can't see that its effort and where the weight actually sits point in different directions.

AI search has a measurement problem. A customer asks ChatGPT what to buy, gets the answer, then comes to your site through Direct, branded search, or another path. Your analytics sees the click. It just doesn’t see AI’s role in creating it. Across the data we’ve analyzed, clicks represent less than 13% of the true impact AI search drives. Today, we’re changing that. We’re releasing Revenue Attribution for AI Search in Goodie, the first feature of its kind in the category. You’ll now see two numbers: 1. Attributed: Revenue from AI clicks your analytics can already identify. 2. Influenced: The zero-click demand AI helped create, but that gets buried in Direct, Organic, branded search, or goes unattributed entirely. Goodie models that influenced revenue using signals across the platform, then triangulates it against post-conversion surveys. So instead of asking whether AEO is actually driving revenue, you can finally measure how much. Visibility and citations matter, but eventually every serious AEO program has to answer one question: Did it drive the business? Now you can. Live today in Goodie. See here for more details: higoodie.com/features/ai-an…

43% of teams call AI search a core strategy in 2026. Only about 14% actually track it. That gap comes down to which metrics teams default to. Traditional analytics assumes a click happened before anything else can be measured. AI search breaks that assumption, because an AI Overview result can answer a buyer's question using your content, without the buyer ever clicking through. GA4 will show nothing, but the content still technically did its job. That's the zero-click problem, and it's why visibility metrics like citation rate have to come before traffic metrics. Traffic can't tell you what it doesn't know happened.

There's a reason we call SEO, AEO, AIO, and GEO the “search quartet”. Each layer makes the next more effective, and brands that treat them as separate work streams leave that compounding effect on the table. SEO, AEO, and AIO draw from the same content work, structure, sourcing, and clarity. Fix one, and the other two improve alongside it, since the underlying signal search engines and AI models read is the same signal. GEO sits apart. It needs that foundation to already exist, because the trust required for an AI agent to act on your behalf gets built through that content work first. Skip the foundation and fund GEO anyway, and the investment has nothing to run on. Build the quartet together, and each layer starts reinforcing the others instead of competing for the same budget line.

keyboard_arrow_left Previous keyboard_arrow_right Next

Most Reddit strategies are built around one question: what will people engage with? But a thread can win every upvote in its subreddit and never once get pulled into an AI answer. This has nothing to do with quality. It’s about portability. Content needs to be built to be lifted out of its original thread and still make sense on its own. Content built for the moment it happened isn't content AI can use, no matter how well that moment landed. Plan for both. Just plan for them separately.

keyboard_arrow_left Previous keyboard_arrow_right Next