Track AI visibility with demand-driven prompts
Build 20+ prompts from seeds with real demand signals, then run them on nine AI platforms.
Prompt for your agent
The ready-to-paste prompt that walks your agent through this work is free with an account.
Create a free accountWhy this matters
Build at least 20 prompt seeds from demand signals such as search volume, Search Console impressions, first-party data or engagement, then rewrite them as natural-language prompts with distinct intent. Run every prompt on each major AI platform with memory disabled, and record the answers and citations so AI visibility can be compared over time.
Before you start
- Access to demand data: Search Console, a keyword volume tool, or first-party analytics
- A dedicated email address for AI platform test accounts
- Accounts on the AI platforms being tested
How to apply it
Collect at least 20 seed topics with demand signals
Pull candidate topics from sources that already show demand: keyword search volume, Search Console impressions, first-party query or sales data, or on-site engagement. Keep only topics whose intent differs from the others so the final prompt set is not a list of near-duplicates.
Check
A list of 20 or more seed topics exists, each paired with the demand signal that justified it (for example a Search Console query and its impressions).
Convert the seeds into evidence-based prompts
Run the seeds through the author's conversion prompt, which turns the topic list into natural-language prompts for AI visibility tracking and enforces a different intent per prompt.
Check
The resulting prompt set contains 20 or more natural-language questions and a review finds no two prompts targeting the same intent.
Set up isolated test accounts on each platform
Create a dedicated email address for AI testing and register accounts on the platforms to be tracked: Google AI Overviews, Google AI Mode, Gemini, ChatGPT, Claude, Perplexity, Grok, Meta and Deepseek.
Check
You can sign in on every listed platform using only the dedicated test address.
Turn off memory on every account
Disable memory or chat-history personalization in each account's settings before running any prompt, so previous sessions do not influence the answers or the citations.
Check
Each account's settings page shows memory or chat history personalization switched off.
Run every prompt on every platform and log answers with citationsGeneral best practice
Execute each prompt on each platform and record the answer text plus every URL it cites. Keep the wording of each prompt fixed from run to run so changes reflect platform behaviour, not prompt edits.
Check
A log contains one row per prompt-platform pair with the answer, the cited URLs and the run date.
Re-run the same prompt set on a fixed cadenceGeneral best practice
Schedule repeat runs (for example monthly) of the unchanged prompt set to build a baseline and detect gains or losses in citations over time.
Check
Two logged runs exist for the same prompt_id with different run_date values and identical prompt_text.
This happens outside your codebase
Work through the steps above on the named site or dashboard, then record the result in your agent session. There is no repository edit for this tip.
Read the source post
https://x.com/i/article/2111887719005667328 AEO case study #1 - 1000% more visibility (in less than 30 days) It's September 1st... And I have to drive sales for the AI SEO for Dummies launch in 21 days. The situation: ChatGPT and AI platforms are NOT recommending the book when you run prompts like: "What are the best AI SEO books for beginners?" "What books should an agency owner give their team to learn AI SEO?" "What are the best AEO books right now?" The weird part is that I did this on purpose. My goal was simple: Prove that you don't need 6-12 months (like SEO) to get AI to recommend your products or services. So here's what I did: ## 1. Create Evidence Based Prompts Any dork can ask ChatGPT to create prompts for them in two seconds. The smarter thing to do is use seeds that have demand signals. Demand signals = search volume, impressions, first party data, user engagement, etc Identify at least 20 seeds with unique intent. Then use the following prompt: Convert these topics into evidence based prompts that use natural language for AI visibility tracking. Important: every prompt should have different intent, use cases, etc. [TOPICS] Now you'll run these pr
seo-mcpreadyCopy the prompt for your agent
Or call it directly
get_tip("track-ai-visibility-with-demand-driven-prompts")Free with an account. Let your agent apply this tip and 175 more.
Connect your agentLet your agent apply this + 175 more tips
Every tip carries the exact steps, a verify command and its source.