When an AI responds to a prompt, it doesn't stop at one answer. It branches into 3–5 follow-up questions. Hema captures all of them. Here's what Query Fanout is, why traditional SEO has no equivalent, and how to use the word-level diff to find content gaps before your competitors do.
The limitation of tracking a single prompt
Most AEO tools — including earlier versions of Hema — track one prompt at a time. You type in "best AEO tracking tool" and see where your brand ranks in ChatGPT's response. That's useful. But it misses something fundamental about how AI search works.
When a user asks ChatGPT a question, the model doesn't just answer it once and stop. It generates a response and then — in most interfaces and in most model architectures — it immediately branches into related questions. These are the "People Also Ask" equivalents of AI search. Except instead of appearing in a sidebar from historical search data, they're generated in real time by the same model that just answered your question.
This branching is called Query Fanout. And it matters more than most AEO practitioners realise.
What Query Fanout actually is
Query Fanout is the set of follow-up questions that an AI model generates after responding to a prompt. For every prompt run Hema executes, the platform captures up to 5 of these AI-generated follow-up questions — alongside the original AI answer.
These five questions are not random. They are what the AI model predicts a curious user would want to know next — based on its training on millions of real user conversations. They are, in effect, the most natural follow-up questions from people who have already asked your target prompt. That makes them a uniquely high-value signal: they're the content gaps your audience will hit immediately after their first AI search.
Reading the word-level diff
Hema doesn't just show you the fanout questions — it shows you how those questions change over time. The word-level diff highlights:
- •Added terms (green) — words that appeared in fanout questions this week but not last week. These are emerging topics in your category.
- •Dropped terms (red) — words that disappeared. These are fading topics you can deprioritise.
- •Kept terms (grey) — stable, persistent topics that define your category's core questions.
In this example, the fanout diff tells a content team: the AI is increasingly associating agency AEO tools with white-labelling (new term) and with LLM-specific comparisons (new term). The term "Citation Share" is emerging alongside "Share of Voice" — suggesting the AI is developing more nuanced distinctions between the two metrics. A content team that spots this in week 2 can publish on "Citation Share vs Share of Voice" before competitors even notice the trend.
3 worked examples
SaaS comparison query
Prompt: "best project management tool for remote teams"
Fanout questions generated by ChatGPT: "Does Stackflow integrate with Slack and Notion?", "What is the difference between Stackflow and Asana for async teams?", "How does Stackflow handle time zone differences in team projects?", "Is Stackflow GDPR compliant for EU remote teams?", "How does Stackflow pricing compare per seat?"
What this tells the content team: Integration questions, competitor comparisons, time zone/async workflows, GDPR compliance, and per-seat pricing are the five content areas most likely to come up immediately after the main query. If Stackflow doesn't have dedicated pages for each of these, they're leaving content gaps that competitors can fill.
Healthcare local query
Prompt: "best cardiologist in Manchester"
Fanout questions: "How to check a cardiologist's NHS credentials?", "What is the difference between a cardiologist and a cardiac surgeon?", "How long is a cardiology referral wait time in Manchester?", "Does [Clinic Name] accept private health insurance?", "What are the signs you need to see a cardiologist urgently?"
What this tells the clinic: Patients are immediately asking about credentials, wait times, insurance, and urgency signals. A clinic with FAQ content answering all five of these questions will be cited more often than a clinic whose website only contains a consultant bio and a contact form.
E-commerce product query
Prompt: "best waterproof running shoes under £120"
Fanout questions: "What is the difference between waterproof and water-resistant running shoes?", "Are Gore-Tex running shoes worth the price?", "Can you run in waterproof shoes in summer without overheating?", "What brands make the most breathable waterproof trail shoes?", "How long do waterproof running shoes last before the membrane fails?"
What this tells the retailer: Buyers are immediately asking about material technology (waterproof vs water-resistant, Gore-Tex), breathability concerns, brand comparisons, and durability. A retailer with a "Waterproof Running Shoes Buying Guide" covering all five points will appear in the follow-up answers, not just the original recommendation.
How to use fanout data for your content calendar
Query Fanout data answers one of the hardest questions in content marketing: what should I write next? Instead of guessing, you have a data-driven answer from AI itself.
- •Step 1: Identify your top 10 tracked prompts. In Hema, go to Prompts → Your Prompts. Sort by Visibility Score (ascending) to find the prompts where you're most invisible.
- •Step 2: Review the fanout questions for each prompt. Open each prompt and click "Query Fanout." Read the 5 follow-up questions. Highlight any question your site doesn't currently answer.
- •Step 3: Look at the word-level diff. Check the diff tab for any new terms that have appeared in the last 2–4 weeks. New terms = emerging topics.
- •Step 4: Prioritise your content calendar. Write articles that directly answer the most frequently appearing fanout questions across your top 10 prompts.
- •Step 5: Track the impact. After publishing, watch whether the fanout questions themselves change. If AI starts generating fanout questions that reference your new article's topic, you've successfully shifted the conversation.
Query Fanout is the only metric in AEO that shows you not just where you are in AI search — but where the conversation is heading. Every competitor tracking only rank and mention rate is looking at a rear-view mirror. Fanout data is the windshield.