Companion piece to Why AI Business Accuracy matters. That post explains why AI can misrepresent a business. This one explains why the specific fixes depend on what kind of business you run.
The foundational research on AI visibility is a 2024 Princeton paper called GEO: Generative Engine Optimization. Aggarwal, Murahari, Rajpurohit, Kalyan, Narasimhan, and Deshpande built the first benchmark for AI-answer visibility and showed that specific content changes — adding cited statistics, quoting authoritative sources, using precise technical language — raised what they call impression score by up to 40% on the same underlying query. (Aggarwal et al., 2024)
That paper established that AI can be moved.
What it did not establish is who has to do what.
The pattern the research points to
Different AI engines source from different corners of the internet. ZipTie's cross-platform citation analysis found that only about 11% of domains appear in both ChatGPT and Perplexity for the same query — roughly 71% appear on only one platform or the other. (ZipTie: How Different AI Platforms Cite the Same Source Differently)
The larger Profound study of 30 million citations across ChatGPT, Google AI Overviews, and Perplexity between August 2024 and June 2025 confirmed the pattern. Each engine has favorites. ChatGPT leans on Wikipedia (47.9% of top citations). Perplexity leans on Reddit (46.7%). Google's AI Overviews pull heavily from both YouTube and Reddit, with Yelp and Facebook showing up in more recent Peec AI analysis. Claude leans on long-form blogs (43.8%). (Search Engine Roundtable summary of Profound data; Search Engine Land: AI Search Engines Cite Reddit, YouTube, and LinkedIn Most)
Yext's analysis of 6.8 million AI citations added a second layer, and it is the one that matters most for small business owners. The mix depends on what kind of question the buyer asked. Factual queries — "what does this company do," "how much does this cost," "where is it located" — get answered mostly from brand-controlled sources: the business's own website and its business listings. That accounts for about 86% of citations. Opinion queries — "which one is best," "who should I hire," "is this any good" — flip almost entirely the other way, to Reddit, YouTube reviews, and community threads. (Surfer: How LLM Citations Work)
Different customers ask different kinds of questions.
Which means different businesses need different sources.
Three playbooks, one rubric
Local businesses live in opinion territory. A coffee shop's buyer is asking best latte in town, cutest brunch spot, good place to bring a first date. These are subjective, community-anchored queries. Google Business Profile, Yelp reviews, Reddit threads about the city, and a handful of local blog mentions do more than a perfectly-tuned website. This is not theoretical. In February 2026, Yelp signed a formal partnership with OpenAI giving ChatGPT direct access to its review corpus, explicitly because roughly half of all search queries carry local intent and OpenAI needed Yelp's content to answer them. (Diginomica: Yelp's Search tie-up with OpenAI) ChatGPT itself confirms it searches the live web through Bing to bring in fresh local information. (OpenAI: ChatGPT search)
Services businesses live in advice territory. A coach or consultant's buyer is asking how do I switch careers, what does executive coaching actually cost, do I need a bookkeeper yet. These are advice queries, and the source pattern is different again. LinkedIn ranks as one of the most-cited domains for professional queries in AI search, alongside long-form blog posts, podcast appearances, and being quoted in credible third-party publications. (Search Engine Land: AI Search Engines Cite Reddit, YouTube, and LinkedIn Most) A Google Business Profile that a coach cannot even qualify for is not the lever here. Thought leadership placed where AI already looks is.
Product and SaaS businesses live in comparison territory. A software buyer is asking best CRM for a two-person team, cheapest scheduling tool with a good API. These are structured comparison queries. G2, Capterra, third-party comparison articles, and clean factual pages on the product's own site — with pricing, integrations, and use cases explicit — carry the day. Pouring energy into Reddit is unlikely to move the needle for a factual SaaS query the way it does for a coffee shop's opinion query.
Same underlying research. Three distinctly different playbooks.
Why this matters for measurement
A generic AI visibility score treats these three businesses as interchangeable. They are not.
A local business scoring poorly on LinkedIn presence is answering the wrong question. A SaaS company scoring poorly on Yelp is answering the wrong question. A consultant scoring poorly on Google Business Profile is answering the wrong question the platform will not even let them attempt.
The rubric has to be the same across businesses so results are comparable. The recommendations cannot be.
Welcome to Daizie AI visibility!
Three playbooks is more than most small business owners want to figure out on a Tuesday. Which one is yours? Which specific moves would matter most, in which order, for the business you actually run?
That is exactly the question Daizie is built to answer.
Daizie is an AI visibility assessment for small businesses. Same research-backed rubric for every business. The recommendations that come out — the specific readiness fixes, the specific accuracy fixes — are calibrated to whether you're local, services, or product. So the coffee shop gets a Yelp and Google Business Profile roadmap. The consultant gets a LinkedIn and thought-leadership roadmap. The SaaS company gets a G2, Capterra, and comparison-content roadmap. Same rubric. Three different playbooks. One assessment that meets you where you actually are.
If you are a little dazed after this post, that is exactly the space Daizie was built for. And if you want to bloom in the answers that matter to your specific kind of business, that is the place to start.
— Marty
