Tracked questions are the foundation of everything you see in QuerySignal. Every metric  visibility, position, citations, sentiment is derived from running these questions against AI platforms daily. Choosing the right ones determines whether your dashboard reflects real market insight or noise.

This guide walks you through how to think about topics, what makes a good tracked question, and how to refine your set over time.

What Are Topics and Tracked Questions?

Topics are high-level themes that organize your tracked questions into meaningful groups. Think of them as the strategic pillars of your brand's presence in AI answers.

Tracked questions are the actual prompts that QuerySignal sends to AI platforms (ChatGPT, Gemini, Claude, Perplexity etc) every day. Each question runs once per day on each platform, and the AI's response is analyzed for brand mentions, competitor mentions, citations, sentiment, and position.

Your configuration includes tracked questions across topics, which are run daily. Results feed every metric and chart on your dashboard.

Writing your Topics and Questions

Step 1: Start With Your Topics

During onboarding, QuerySignal analyzes your website and generates up to 10 suggested topics for your brand. You can select from these or add your own.

Example Good topics cover:

Tips for choosing topics:

• Aim for 5-10 topics to start broad enough that each one can support multiple questions

• Think about what your ideal customer would ask an AI assistant about your category

• Include at least one competitor comparison topic QuerySignal auto-generates a "[Brand Name] vs Competitors" topic for this reason

• Each topic should represent a distinct area of your business or market

Step 2: Write Effective Tracked Questions

Once your topics are defined, populate each with questions that reflect how real people ask AI assistants for help. This is the most important step.

What makes a good tracked question:

Examples:

Cover different intent types to get a complete picture of your AI presence:

1- Informational (Awareness Stage)

These are early-stage questions. The user is learning about the category.

SaaS example

• "What is customer data platform software?"

• "How does cloud backup work for small businesses?"

Ecommerce example

• "What are the benefits of organic skincare products?"

• "What is the difference between whey and plant-based protein?"

B2B services example

• "What does a fractional CFO do?"

• "How does cybersecurity risk assessment work?"

These questions tell you whether your brand appears when people first discover your category.

2- Comparative (Consideration Stage)

Here the user is evaluating options.

SaaS example

• "Best HR software for companies under 200 employees"

• "Top CRM tools compared for startups"

Ecommerce example

• "Best noise-canceling headphones under $300"

• "Nike vs Adidas running shoes comparison"

B2B services example

• "Top digital marketing agencies for SaaS companies"

• "[Your Brand] vs [Competitor]"

These are high-value questions. If AI consistently mentions competitors but not you, this is a direct visibility gap.

3- Problem-Solving (Action Stage)

The user has a clear problem and wants a solution.

SaaS example

• "How do I automate invoice reminders?"

• "How can I reduce churn in a subscription business?"

Ecommerce example

• "How do I fix dry skin in winter?"

• "Best way to improve home WiFi coverage"

B2B services example

• "How can I reduce cloud infrastructure costs?"

• "How do I improve LinkedIn lead generation?"

These questions often convert well because they reflect real operational pain.

4- Transactional (Purchase Stage)

The user is ready to buy or shortlist vendors.

SaaS example

• "Affordable payroll software for small businesses"

• "Enterprise project management software pricing"

Ecommerce example

• "Where can I buy halal protein powder online?"

• "Discount codes for premium office chairs"

B2B services example

• "Hire a cybersecurity consultant in London"

• "Best ERP implementation partner for manufacturing"

Step 3: Use Your Own Customer Data

Your support tickets, sales calls, and FAQs are a goldmine for tracked questions. These reflect how your actual customers talk and what they care about.

How to translate internal data into tracked questions:

These problem-driven questions tend to perform well because they reflect real language and real context.

Step 4: Generate Questions With AI

If you need help expanding your question set, use the Generate feature in Tracked Questions:

1. Go to Tracked Questions and click Modify Questions

1. Click Generate

1. Select which topics to generate questions for

1. Optionally add custom instructions to guide the generation (e.g., "focus on enterprise buyers" or "include pricing-related questions")

1. Review the suggested questions and add the ones that fit

You can also generate entirely new topics if you want to explore adjacent areas of your market.

How Many Questions Should I Track?

There is no single right answer, but here are guidelines:

• Start with 5-10 questions per topic : enough to see patterns, not so many that you're overwhelmed

• Aim for 25-50 total questions to get a meaningful picture of your AI presence

• Quality over quantity : 50 well-chosen questions will give you better insights than 200 generic ones

Your plan determines the maximum number of questions you can track. You'll see a notification if you exceed the limit.

Refining Your Questions Over Time

Your question set shouldn't be static. Revisit it monthly:

• Check Weak Prompts : questions where your visibility is below 50% but competitors are mentioned. These are the battleground questions worth keeping.

• Remove low-value questions : if a question consistently returns irrelevant results or no competitor mentions, consider replacing it

• Add questions for emerging trends : as your market evolves, new topics and questions become relevant

• Look at individual question metrics : click any question to see its visibility, average position, and citation share over time. This tells you whether a question is worth tracking.

You can stop tracking, resume, duplicate, or delete any question from the Modify Questions page without losing historical data.