AI Visibility Metrics: Track Your Brand for Free

A zero-budget workflow to monitor whether ChatGPT, Gemini, Perplexity, and Claude mention your product

Learn how to build a free AI visibility tracking system in under 30 minutes per week. This step-by-step tutorial walks you through querying four major AI platforms, logging brand mention share, and establishing a baseline score — no enterprise tools required.

TL;DR

You can track AI visibility for free - Build a spreadsheet, write 5-10 category prompts, and query ChatGPT, Gemini, Perplexity, and Claude weekly to measure whether AI engines mention your product.
Four metrics matter most - Brand Mention Rate, Brand Mention Share (share of voice), Citation Rate, and Recommendation Rate. Calculate all four from your manual query data.
Run a baseline audit first - Query all prompts across all platforms in one session (about 45 minutes) to establish your starting point. Even 0% visibility is a valid and useful baseline.
Weekly maintenance takes 20-30 minutes - Rotate through your prompt library, log results, update your metrics dashboard, and track drift (changes over time) to spot trends.
Source analysis reveals where to invest - Ask AI platforms what sources informed their responses. The patterns tell you whether to focus on listicle placements, Reddit engagement, comparison posts, or other third-party authority channels.

What You Will Build: A Zero-Budget AI Visibility Tracking System

By the end of this tutorial, you will have a working system to monitor GEO signals and track whether AI engines like ChatGPT, Gemini, Perplexity, and Claude mention your product. No enterprise tools. No monthly subscriptions. Just a repeatable workflow you can run in under 30 minutes per week.

Your success criteria are concrete: you will have a spreadsheet tracking your AI visibility metrics across at least four major AI platforms, a baseline score for your current visibility, and a documented process to measure changes over time. You will know exactly where you stand before spending a single dollar on optimization.

This matters because in AI-driven search, influence replaces traffic as the primary indicator of visibility. If ChatGPT doesn't know your product exists, an entire discovery channel is closed to you.

Prerequisites and Setup Checklist

Before you start, confirm you have these in place. Missing any one of them will stall you mid-process.

A launched product with a live website. Even a landing page counts. AI models need something indexable to reference.
Free accounts on four AI platforms:ChatGPT (free tier), Google Gemini, Perplexity, and Claude. Create these now if you haven't.
A Google Sheet or Notion database. You need somewhere to log results consistently.
A list of 5-10 prompts your ideal customer would type. Think category queries, not brand queries. Example: "best budget project management tool for freelancers" rather than "what is [YourBrand]."
30 minutes per week. This is the ongoing time commitment.

Time estimate for initial setup: 60-90 minutes. Weekly maintenance: 20-30 minutes.

Why This Manual Approach Works for Pre-$1k MRR Founders

Enterprise GEO platforms like Rankfender, Otterly, or Peec AI charge $100-500+ per month. They are built for marketing teams tracking hundreds of keywords across dozens of competitors. If you are pre-$1k MRR, that math does not work.

The manual method described here gives you the same core signal: does AI know my product exists, and is it recommending me? You lose automation and historical dashboards. You gain the ability to start today, for free, and build the muscle of understanding exactly how AI engines talk about your category.

This approach is not a permanent solution. It is a bridge. Once you cross $1k MRR, you can graduate to paid tools with the baseline data you have already collected. Until then, every dollar matters more than every dashboard.

Step 1: Build Your Prompt Library

Open a new Google Sheet. Create a tab called "Prompts" with three columns: Prompt, Category, and Intent Type.

Write 5-10 prompts that a potential customer would ask an AI assistant when looking for a product like yours. These should be category-level, problem-aware queries. Avoid branded queries for now.

Examples for a task management SaaS:

Prompt: "What are the best simple task management apps for solo founders?"
Category: Task Management
Intent Type: Discovery
Prompt: "Recommend a lightweight project tool under $20/month"
Category: Task Management
Intent Type: Purchase
Prompt: "Alternatives to Trello for one-person teams"
Category: Task Management
Intent Type: Comparison

Checkpoint: You should have at least 5 prompts covering discovery, comparison, and purchase intent. If you are stuck, search Reddit or Twitter for how people describe the problem your product solves, and convert those phrases into AI prompts.

Common failure: Writing prompts that are too specific to your product name. AI engines will not mention you if you ask about you. Test category queries first to see if you appear organically.

Step 2: Create Your Tracking Spreadsheet

In the same Google Sheet, create a second tab called "AI Visibility Log." Set up these columns:

Date | Platform | Prompt Used | Mentioned? (Y/N) | Position (1st/2nd/3rd/Not Listed) | Exact Quote | Competitors Mentioned | Link Cited? (Y/N) | Sentiment (Positive/Neutral/Negative)

This structure captures the core AI visibility metrics that matter: Brand Mention Rate, Share of Voice, Citation Rate, and Recommendation Rate. You are tracking all four manually.

Checkpoint: Your spreadsheet should have clear headers and be ready to receive data. Add a frozen header row so it stays visible as the sheet grows.

Common failure: Skipping the "Competitors Mentioned" column. This is how you calculate your brand mention share, which is the percentage of mentions you receive versus competitors in the same response. Without it, you have data but no context.

Step 3: Run Your First Baseline Audit

Open all four AI platforms in separate browser tabs. Take your first prompt and paste it into each one. Record every result in your tracking spreadsheet.

For each response, capture:

Whether your product was mentioned at all
Where in the response it appeared (first recommendation, middle of a list, or absent)
The exact sentence where your product was mentioned (copy-paste it)
Every competitor name that appeared in the same response
Whether the AI linked to your website
Whether the tone was positive, neutral, or negative

Repeat this for every prompt in your library, across all four platforms. Yes, this means 20-40 individual queries for your first baseline. It takes about 45 minutes.

Checkpoint: Your spreadsheet should now have 20-40 rows of data. If your product was mentioned zero times, that is a valid and important baseline. Do not skip this step or assume the answer.

Common failure: Running prompts while logged into accounts that have previous context about your product. Use incognito or private browsing windows to get unbiased results.

Step 4: Calculate Your Baseline Metrics

Create a third tab called "Metrics Dashboard." Here you will compute four numbers from your raw data.

Brand Mention Rate: Count the total responses where your product was mentioned. Divide by total queries run. Multiply by 100. This is your mention percentage.

Brand Mention Rate = (Responses mentioning you / Total queries) × 100
Example: 3 mentions out of 40 queries = 7.5% mention rate

Brand Mention Share (Share of Voice): Count every competitor mention across all responses. Add your own mentions. Your share is your mentions divided by total mentions (yours + competitors).

Brand Mention Share = (Your mentions / All brand mentions in responses) × 100
Example: 3 your mentions, 47 total mentions = 6.4% share of voice

Citation Rate: Of the responses that mentioned you, how many included a link to your website?

Recommendation Rate: Of the responses that mentioned you, how many positioned you as a recommended or suggested option (versus just listing you)?

According to AIROPS research on AI visibility metrics, these five signals (citation share, competitive share of voice, mention rate, sentiment score, and drift) are the clearest indicators of brand performance in AI search.

Checkpoint: You now have four numbers. Write them down with today's date. These are your baseline. Everything you do from here is measured against them.

Step 5: Set Up Weekly Monitoring

Pick one day per week. Tuesday or Wednesday works well because AI model updates often roll out early in the week.

Each week, run 3-5 of your prompts across all four platforms. Rotate which prompts you use so you cover your full library every two weeks. Log everything in your tracking spreadsheet using the same format.

Weekly time commitment: 20-30 minutes.

After each session, update your Metrics Dashboard tab. The critical metric to watch is drift, which is the change in your visibility scores over time. A founder who was invisible last month but appears in 15% of queries this month is making real progress.

If you are already using a daily growth loop to capture intent signals from user behavior, add this weekly AI visibility check as a standing item. It takes the same operational discipline.

Common failure: Doing this enthusiastically for two weeks, then forgetting. Set a recurring calendar event. Treat it like checking your Stripe dashboard.

Step 6: Add Google AI Overview Tracking

AI Overviews (the AI-generated summaries at the top of Google search results) are a separate visibility channel. You should track them independently.

Take your 5-10 prompts and rephrase them as Google search queries. Search each one in Google (logged out, incognito) and note:

Did an AI Overview appear for this query?
Were you mentioned in the AI Overview?
Were competitors mentioned?

According to VisibilityStack AI research, brands should track the percentage of keywords triggering AI Overviews and their inclusion rate within those overviews separately from chatbot visibility.

Add a column to your spreadsheet for "AIO Triggered (Y/N)" and "AIO Mentioned (Y/N)." This takes an extra 10 minutes per weekly session.

Checkpoint: You now have visibility data across five channels: ChatGPT, Gemini, Perplexity, Claude, and Google AI Overviews.

Step 7: Monitor Brand Mentions With Free Alerts

Manual querying catches AI responses. But you also need to know when your brand appears in the sources AI models pull from (blogs, forums, comparison sites, listicles).

Set up these free monitoring tools:

Google Alerts: Create alerts for your brand name, your founder name, and your product category + "best" or "alternatives." Set delivery to "as it happens."
Talkwalker Alerts (free): A second alert system that catches mentions Google Alerts misses. Set up identical queries.
Reddit search: Bookmark a Reddit search URL for your brand name and check it weekly. Reddit threads are heavily cited by AI models.

These alerts feed into your understanding of third-party authority and content citability. When an AI engine mentions you, it is almost always because a third-party source mentioned you first. Tracking those sources tells you where to focus your efforts.

Common failure: Setting alerts that are too broad. "Project management" will flood your inbox. Use specific phrases like "[YourBrand] review" or "[YourBrand] vs."

Step 8: Identify What Drives Mentions (Source Analysis)

After 2-3 weeks of data collection, look at the responses where you were mentioned. Ask each AI platform: "What sources did you use for that recommendation?" Perplexity shows sources automatically. For ChatGPT and Claude, follow up with: "Can you share the sources that informed your response about [your product]?"

Record these sources in a new tab called "Source Analysis." You will start seeing patterns: maybe your Product Hunt launch page gets cited, or a blog post comparing you to competitors, or a specific Reddit thread.

This tells you where to invest your limited time. If listicle placements drive AI mentions, write to listicle authors. If Reddit threads drive mentions, engage authentically on Reddit. If nothing drives mentions yet, you know you need to build third-party authority from scratch.

For founders still figuring out which channel deserves their limited time, these seven diagnostic signals can help you prioritize.

Configuration and Customization

Variables You Should Adjust

Number of prompts: Start with 5-10. If your product spans multiple categories, expand to 15-20 over time. More prompts give you better coverage but take more time.

Platform selection: The four platforms listed (ChatGPT, Gemini, Perplexity, Claude) cover the majority of AI search traffic. If your audience skews technical, add DeepSeek. If they skew toward Twitter/X users, add Grok.

Tracking frequency: Weekly is the sweet spot for pre-$1k MRR. Daily is overkill (AI models do not update that fast). Monthly is too slow to catch trends.

Settings You Must Change

Always use incognito/private browsing. AI platforms personalize responses based on your history. Your logged-in results will be skewed.

Rotate prompt phrasing slightly each month. AI models can return different results for "best task app for freelancers" versus "top task management tools for solo workers." Test variations to avoid blind spots.

Tools like heycatch can help automate parts of this discovery process by surfacing competitor positioning and identifying which growth actions to prioritize each day, which is useful when you are ready to act on the data you are collecting here.

Verification and Testing

Your system is working correctly if you can answer these three questions after your third weekly session:

What is my current Brand Mention Rate across AI platforms? (A specific number, even if it is 0%.)
What is my Brand Mention Share relative to competitors? (A specific percentage.)
Is my visibility trending up, down, or flat? (Requires at least 3 data points.)

Edge cases to verify: run one branded query ("What is [YourProduct]?") to confirm AI engines have basic awareness of your product. If they return nothing or incorrect information, that is a signal to focus on building foundational third-party mentions before worrying about category-level visibility.

Also test one prompt that is adjacent to your category but not directly in it. This reveals whether AI engines associate you with related topics, which expands your discoverability surface.

Common Errors and Fixes for Monitoring AI Visibility Metrics

Error: "AI gave me completely different results today than last week"

Cause: AI models are non-deterministic. The same prompt can produce different outputs on different days. Fix: This is expected behavior, not a bug. Run each prompt 2-3 times per session and record the most common result. GetFancy.ai recommends 60-100 prompt repetitions for statistical reliability, but 2-3 repetitions per prompt is a practical minimum for solo founders.

Error: "My product is never mentioned in any response"

Cause: Your product lacks sufficient third-party mentions for AI models to reference. Fix: Focus on getting mentioned in 3-5 external sources first (comparison blog posts, Reddit threads, Product Hunt, indie hacker communities). AI models cite what the web already says about you.

Error: "Perplexity mentions me but ChatGPT does not"

Cause: Different AI models have different training data cutoffs and retrieval methods. Perplexity searches the live web; ChatGPT relies more on training data. Fix: Track each platform separately. Optimize for Perplexity first (it is the easiest to influence because it uses real-time web retrieval), then work on building the persistent web presence that ChatGPT and Claude pick up during training updates.

Error: "I set up Google Alerts but get zero notifications"

Cause: Your brand name is too generic or too new. Fix: Add context to your alerts. Instead of just "TaskFlow," try "TaskFlow app" or "TaskFlow project management." Also set up alerts for your exact domain name.

Error: "My spreadsheet is getting messy and hard to read"

Cause: No data validation or formatting rules. Fix: Use Google Sheets data validation dropdowns for Y/N columns and sentiment columns. Add conditional formatting to highlight rows where you were mentioned (green) versus not mentioned (red). This takes 5 minutes and saves hours of scanning.

Next Steps and Extensions

Once you have 4-6 weeks of data, you are ready to move from measurement to action. Here are three ways to extend this work:

Content citability audit: Review your website and blog content through the lens of AI citation. Does your content contain clear, quotable statements? Does it use structured data and schema markup? AI models prefer content that is easy to extract and attribute.
Competitor gap analysis: Use your "Competitors Mentioned" data to identify which competitors appear most often and analyze why. What sources do they appear in that you do not? Target those sources.
Separate vanity metrics from real demand signals. Your AI visibility data should connect to actual user acquisition. If mentions go up but sign-ups stay flat, you may be optimizing the wrong prompts. This framework for separating revenue signals from noise can help you connect the dots.

The goal is not to live in a spreadsheet forever. The goal is to build enough signal to know where to invest, whether that is Generative Engine Optimization, digital PR for AI, or doubling down on the channels already working. Measure first. Then act with confidence.

Frequently Asked Questions

What is AI Search Visibility and why is it important?

AI visibility is the quantitative measure of how often and how prominently a brand appears in responses generated by AI assistants and generative search engines. It matters because more people are using ChatGPT, Gemini, and Perplexity to discover products instead of traditional Google searches. If AI engines do not know your product exists, you are invisible to a growing segment of potential users.

When should I start measuring my AI visibility?

Start the week you launch. Even a 0% baseline is valuable data. Knowing that no AI engine mentions you gives you a clear starting point and motivates action. Waiting until you "have more traction" means you miss the window to understand what drove early changes in visibility.

Which platforms should I monitor for AI visibility?

At minimum, track ChatGPT, Google Gemini, Perplexity, and Claude. These cover the largest share of AI-assisted search. Add Google AI Overviews as a fifth channel since they appear directly in search results. If your audience is technical, consider adding DeepSeek. If they are active on X/Twitter, add Grok.

How does Generative Engine Optimization (GEO) improve brand visibility?

GEO focuses on making your content and brand more likely to be cited, summarized, or recommended by AI models. This includes building third-party mentions on sites AI models trust, creating structured and quotable content, and ensuring cross-source consistency in how your product is described. Unlike traditional SEO, GEO prioritizes being recommended inside AI answers rather than ranking on a results page.

Can I track AI visibility without paying for tools?

Yes. The manual method in this tutorial covers the same core metrics (mention rate, share of voice, citation rate, recommendation rate) that paid platforms track. You trade automation for cost savings. For solo founders under $1k MRR, the manual approach provides enough signal to make informed decisions without adding a monthly expense.

How often do AI models update their knowledge of new products?

It varies by platform. Perplexity searches the live web with each query, so new mentions can appear within days. ChatGPT and Claude rely on periodic training data updates, which can lag by weeks or months. Google Gemini falls somewhere in between. This is why tracking each platform separately matters: your visibility timeline will differ across them.