Tutorial: Setting Up Citation Tracking for AI Search — Manual and Automated Approaches

If you cannot measure it, you cannot improve it. Citation tracking is the measurement foundation of GEO -- it tells you which AI platforms cite your content, how often, for which queries, and how that changes over time. Yet only 22% of marketers currently track AI visibility at all, leaving the majority flying blind in a channel where traffic converts at 14.2% compared to Google organic's 2.8%. This tutorial walks you through setting up citation tracking from scratch, starting with a zero-cost manual approach and progressing to automated tools.

What You Are Tracking

Before setting up any system, be clear about what you are measuring. Citation tracking captures four distinct types of AI references:

URL citations are the strongest signal. The AI platform links to a specific page on your site as a source. Perplexity provides numbered source links for nearly every response. ChatGPT Search provides inline citations for real-time queries. These drive actual traffic.

Brand mentions occur when the AI names your brand without linking to you. "Allbirds makes great sustainable sneakers" is a mention. It builds awareness but does not drive clicks.

Product recommendations occur when the AI specifically recommends your product for a user's need. These carry the highest purchase intent.

Indirect citations occur when a third-party source (a review site, a Reddit thread) that mentions your brand gets cited by the AI. Your brand appears in the response, but the traffic goes to the third party.

A complete tracking system captures all four types.

Part 1: Manual Citation Tracking

Manual tracking is where every brand should start. It costs nothing, builds intuition about how AI platforms work, and provides the baseline data needed to evaluate automated tools.

Step 1: Build Your Prompt Library

Create a spreadsheet with 20-30 prompts organized into four categories:

Category prompts (8-10):

  • "Best [your product category] in 2026"
  • "Top [product type] for [use case 1]"
  • "Top [product type] for [use case 2]"
  • "Affordable [product category] recommendations"
  • "Premium [product category] worth the investment"
  • "[Product category] for beginners"
  • "Most popular [product category] brands"
  • "Where to buy [product category] online"

Comparison prompts (5-7):

  • "[Your brand] vs [competitor A]"
  • "[Your brand] vs [competitor B]"
  • "[Competitor A] vs [competitor B]" (track even without your brand)
  • "Compare [product category] options under $[price]"
  • "[Your product] vs [competitor product] for [specific use]"

Problem-solution prompts (4-5):

  • "I need help choosing [product category] for [specific need]"
  • "What [product type] solves [common problem]?"
  • "My [current product] is not working for [use case]. What should I try?"
  • "How to find the right [product] for [specific constraint]"

Brand-specific prompts (3-5):

  • "Tell me about [your brand]"
  • "Is [your brand] good quality?"
  • "What are [your brand]'s most popular products?"
  • "[Your brand] reviews"

Step 2: Set Up Your Tracking Spreadsheet

Create a spreadsheet with these columns:

Column A: Date
Column B: Platform (ChatGPT / Perplexity / AI Overviews / Gemini / Claude)
Column C: Prompt Text
Column D: Brand Mentioned (Yes/No)
Column E: Citation Type (URL / Mention / Recommendation / Indirect / None)
Column F: Cited URL (if applicable)
Column G: Position (1st, 2nd, 3rd mention in response)
Column H: Sentiment (Positive / Neutral / Negative)
Column I: Competitors Mentioned (list)
Column J: Competitor URLs Cited (list)
Column K: Accuracy (Accurate / Inaccurate — note issues)
Column L: Notes

Step 3: Run Your First Tracking Cycle

Set aside three to four hours for your first full tracking cycle. Run each prompt on each platform:

ChatGPT: Use chat.openai.com with the search feature enabled. Ask each prompt and record results. Note that ChatGPT uses Bing's index, so Bing-indexed content is prioritized. Wikipedia is ChatGPT's most cited source at 7.8% of total citations.

Perplexity: Use perplexity.ai. Perplexity provides the most detailed citations with numbered sources. This is the easiest platform for citation tracking because sources are explicitly linked. Reddit is Perplexity's top source at 6.6% of citations.

Google AI Overviews: Search Google for each prompt and note the AI Overview response if one appears. AI Overviews appear in 15-60% of searches depending on query type. Over 91% of ecommerce queries trigger AI results.

Gemini: Use gemini.google.com. Record mentions and any source cards.

Claude: Use claude.ai. Claude mentions brands in 97.3% of answers, the highest rate of any platform, but places mentions later (median rank 3).

Step 4: Run Each Prompt Three Times

AI responses are non-deterministic. The same prompt produces different results each time. Run each prompt three times per platform and record the majority result. If your brand appears in two out of three runs, record it as "mentioned." This accounts for variability and gives you more reliable data.

Step 5: Calculate Your Baseline Metrics

After your first complete cycle, calculate:

Overall Mention Rate: Number of prompt-platform combinations where your brand appeared / total prompt-platform combinations tested. Example: Mentioned in 35 out of 120 tests = 29.2%.

Citation Rate: Number of tests where a specific URL was cited / total tests. This is typically much lower than mention rate.

Platform-Specific Rates: Calculate mention rate for each platform separately. You will likely see significant variation -- Claude mentions brands most (97.3%), ChatGPT moderately (73.6%), AI Overviews least (48.5%).

Share of Voice: Your mentions / (your mentions + competitor mentions) across all tracked prompts.

Step 6: Establish Your Tracking Cadence

Weekly: Run your full prompt library across all platforms. Update your spreadsheet. Calculate weekly metrics and compare to previous week. This takes three to four hours per week.

Monthly: Summarize trends. Calculate month-over-month changes. Identify prompts where visibility improved or declined. Document any pattern changes.

Part 2: Analytics Integration

While manual prompt tracking shows your AI platform visibility, Google Analytics 4 captures the actual traffic and conversions from AI referrals.

Set Up AI Referral Tracking in GA4

Create a custom channel group for AI traffic:

  1. In GA4, go to Admin > Data Display > Channel Groups
  2. Create a new channel group or modify the default
  3. Add a new channel called "AI Search" with these rules:
    • Source matches regex: chatgpt|openai|perplexity|gemini\.google|copilot\.microsoft|claude\.ai
    • OR Medium matches: ai-referral

Create AI-Specific Reports

Build a custom exploration in GA4 that tracks:

  • Sessions by AI referral source
  • Conversion rate by AI platform
  • Revenue attributed to AI referral traffic
  • Pages most visited from AI referrals (these indicate which prompts drive traffic)
  • New vs returning users from AI sources

UTM Parameters for AI Landing Pages

If you create content specifically targeting AI citation (comparison guides, FAQ pages), add UTM parameters to internal links so you can track when AI-driven visitors navigate deeper into your site:

https://www.yourstore.com/products/your-product?utm_source=ai_landing&utm_medium=internal&utm_campaign=geo_optimization

Part 3: Automated Citation Tracking

Once you have baseline data from manual tracking (four to six weeks recommended), evaluate automated tools to scale your monitoring.

When to Move to Automated Tools

Transition to automated tracking when:

  • Manual tracking takes more than four hours per week
  • You need to track more than 30 prompts
  • You need historical trend data with statistical reliability
  • You need competitive benchmarking beyond what manual testing provides
  • You want alerts for significant visibility changes

Setting Up Otterly.AI

Otterly.AI is the most accessible starting point for automated citation tracking:

  1. Create an account and select your plan (Lite at $29/month tracks 15 prompts, Standard at $189/month tracks 100)
  2. Add your domain and brand name
  3. Import your prompt library from your manual tracking spreadsheet
  4. Add competitors for benchmarking (typically three to five)
  5. Configure platforms to monitor (six AI engines included by default)
  6. Set up weekly reports with automated email delivery
  7. Run your first scan and compare results to your manual baseline

The GEO audit tool (included) analyzes 25-plus on-page factors for each tracked URL, providing optimization recommendations alongside tracking data.

Setting Up Semrush AI Toolkit

For teams already using Semrush:

  1. Add the AI Toolkit ($99/month add-on per domain)
  2. Configure brand tracking with your brand name and competitors
  3. Set up AI Share of Voice monitoring for your domain
  4. Import relevant prompts from your manual library
  5. Review the AI perception report to understand how ChatGPT and AI Overviews describe your brand
  6. Set up alerts for significant changes in AI visibility

The advantage of Semrush is integration with existing SEO data, letting you correlate AI citations with traditional search performance.

Setting Up Ahrefs Brand Radar

For larger teams and agencies:

  1. Access Brand Radar (available on the $699/month plan)
  2. Configure brand monitoring across six AI indexes
  3. Leverage the 343 million-plus prompt database for broad coverage beyond your custom prompt library
  4. Set up competitive comparison dashboards
  5. Review historical data for trends predating your monitoring start date (Ahrefs tracks prompts retroactively)

Comparing Automated Outputs to Manual Baseline

After your first automated scan, compare results to your manual tracking:

  • Do the tools report similar mention rates to your manual testing?
  • Are citation patterns consistent?
  • Does the competitive picture match?

Discrepancies may indicate differences in prompt phrasing, timing, or platform behavior. Investigate significant differences to ensure your automated tracking is calibrated correctly.

Part 4: Building Your Citation Dashboard

Combine manual, analytics, and automated data into a single dashboard.

Essential Dashboard Elements

  1. Overall AI Visibility Score — weekly trend with 4-week rolling average
  2. Citation Rate by Platform — bar chart showing ChatGPT, Perplexity, AI Overviews, Gemini, Claude
  3. Share of Voice vs Competitors — stacked bar chart or line chart over time
  4. Top Cited Pages — table showing which of your pages are cited most often
  5. AI Referral Traffic — from GA4, sessions and conversions from AI sources
  6. Revenue Attribution — direct AI revenue plus estimated AI-influenced revenue
  7. Action Items — current optimization priorities based on gap analysis

Dashboard Update Schedule

  • Automated data (tools + GA4): Updates daily or weekly automatically
  • Manual tracking data: Updated weekly
  • Analysis and action items: Updated monthly
  • Executive summary: Prepared quarterly

The Bottom Line

Citation tracking transforms GEO from guesswork into a data-driven discipline. Start with manual tracking to build understanding and baseline data. Add GA4 integration to capture traffic and revenue impact. Graduate to automated tools when manual tracking becomes a bottleneck. The complete system -- manual testing plus analytics plus automated monitoring -- gives you the visibility needed to optimize effectively. With AI-referred traffic converting at five times the rate of organic and only 22% of marketers tracking AI visibility, setting up citation tracking is both a competitive advantage and a business imperative.

Up next

Shopify Schema Setup

Tutorial: implementing schema markup on Shopify.