TLDR: How to Check If AI Cites Your Site in 20 Minutes
You do not need a fancy tool to start. The manual method takes 20 minutes for a first-pass baseline:
- List 10 buyer-intent prompts your customers would type into ChatGPT.
- Run each prompt in ChatGPT (browse enabled), Perplexity, Gemini, and Google AI Overviews.
- Screenshot the responses. Log which domains are cited.
- Repeat monthly. Track the trend.
If any of those 10 prompts is worth real pipeline to you and you are not cited, that is a gap worth fixing. If you want the automated version, tools like Profound and Otterly.ai exist. We cover both below.
Why is checking AI citations different from checking Google rankings?
Google rankings are deterministic. You type a query, get a ranked list, and know exactly where you sit. AI citations are probabilistic. The same prompt in ChatGPT can return different citations depending on time of day, browsing history, user context, and which model version is running.
This means you cannot just check once. You have to sample. Run the same prompt three times in different sessions and average the results. That is closer to the truth than any single query.
The ChatGPT Search release notes and Perplexity's blog both confirm that citation logic uses retrieval-augmented generation with source selection that varies query-to-query.
What are the manual methods to check AI citations?
Four workflows, in order of effort. Start with method 1 for baseline, escalate as you scale.
Method 1: Direct prompt testing
Open ChatGPT with browse enabled, Perplexity, Gemini, and Google Search with AI Overviews. Run each of your 10 target prompts. Screenshot responses. Log cited domains in a spreadsheet.
Example target prompts for a SaaS positioning tool:
- "best SaaS positioning tools for solo founders"
- "how do I fix my SaaS homepage headline"
- "SaaS positioning framework"
- "how to write a better SaaS value proposition"
Time: 20 to 30 minutes for 10 prompts across 4 engines.
Method 2: Brand mention testing
Ask each AI engine directly about your brand: "What is Growth Pigeon?" "Who are the founders of X?" "What products does Y sell?" This tells you whether your brand exists in the model's knowledge at all, and whether the description is accurate.
If ChatGPT says nothing about your brand or gets facts wrong, you have a foundational visibility problem. Fix that before optimizing for target prompts.
Method 3: Competitor citation gap analysis
For every target prompt where a competitor is cited and you are not, note what makes their content different: FAQ sections, use case pages, question-led H2s, off-site LinkedIn presence, Reddit threads. That is your fix list. This is the single highest-leverage manual workflow and it costs nothing except time.
Method 4: Cite-source reverse engineering in Perplexity
Perplexity shows citations inline. Click through each cited source for a target prompt and analyze the on-page pattern. If Perplexity keeps citing FAQ-heavy pages with 20+ external links, you now know the template that wins for that query.
What automated AEO tracking tools exist?
The AEO tools category is early but real. The main players as of mid-2026:
- Profound: enterprise-focused. Tracks brand mentions across ChatGPT, Perplexity, Gemini, Copilot. Reports on competitor citation share.
- Otterly.ai: mid-market. Focuses on citation frequency and sentiment across AI engines. Alerts on new mentions.
- Search Atlas: SEO tool that added AEO tracking. Good if you already use it for SEO.
- Ahrefs: added AI Overview visibility to its rank tracker. Not as deep as Profound but bundled.
- SEMrush: added AI mode tracking. Similar coverage to Ahrefs.
- Brand24 and Mention: general brand monitoring tools that added AI engine coverage.
The category is early. Expect meaningful accuracy improvements every quarter. If you are running AEO seriously, one of these tools is worth $200 to $500 per month. If you are testing the waters, the manual method is fine.
Which prompts should I actually track?
Track prompts, not keywords. The mental shift matters. A keyword is a broad term ("SaaS positioning"). A prompt is a real question ("what is the best SaaS positioning framework for a solo founder").
To generate your prompt list:
- Interview 3 to 5 recent customers. Ask what they typed into ChatGPT before they found you. Their exact wording is gold.
- Check your Search Console for long-tail queries with 4 to 8 words. These are typically prompt-shaped.
- Use Answer The Public to find question-shaped queries in your niche.
- Ask ChatGPT: "What are the top 10 questions someone would ask before buying [your category]?" Then verify with real customer language.
You want 20 to 30 prompts covering the full buyer journey, weighted toward the evaluation stage (55% of AI citations happen there per HubSpot's 2026 research).
How often should I check citations?
Monthly is enough for most SaaS operators. Weekly if you are running an active AEO sprint. Quarterly is too infrequent because AI engines update fast enough that a quarter can hide a citation win or loss.
Sample cadence:
- Monthly: Full sweep across all 20 to 30 target prompts, all 4 engines.
- Weekly: Spot check 5 highest-priority prompts.
- Per-publish: When you ship a new article, check its target prompt weekly for the first month.
What metrics should I report on for AEO?
Track these:
- Citation rate per prompt: percentage of times your domain is cited when a target prompt is run.
- Citation share vs top 3 competitors: your citations divided by your plus top 3 competitor citations. This normalizes for query volatility.
- Brand mention rate: how often your brand is named in AI responses, cited or not.
- New citations per month: how many prompts you newly got cited on.
- Brand search volume: track "your brand name" in Google Search Console. AI-driven discovery shows up here.
Do not track rankings. Rankings are the wrong metric now. Track citations, citation share, and downstream brand demand.
How do I know which fixes actually moved the needle?
Attribution in AEO is hard. Everything is probabilistic. The cleanest workflow:
- Baseline: measure citation rate on 20 prompts across 4 engines.
- Make one specific change: for example, add FAQ schema to 10 articles.
- Wait 4 weeks (Perplexity), 8 weeks (Gemini), 12 weeks (ChatGPT browse).
- Re-measure. Compare citation rate on affected prompts.
Do not make multiple changes at once during test periods. You will not know which one worked. This is slow but it is the only way to build a validated internal playbook.
For a full A/B methodology, see Search Engine Journal's coverage of AEO experimentation or Backlinko's AI Overview studies.
What if my site is not cited at all?
Most SaaS sites are not cited when they first check. That is normal. Common causes and fixes:
- Content is too generic. If your articles are "10 tips for X" listicles, LLMs will not cite them. Rewrite with proprietary data, specific numbers, and unique frameworks.
- No FAQ schema. Add it. +24 lift on AI Overviews. See our FAQ schema for AEO guide.
- Backlink strategy. If you have been chasing backlinks, redirect budget to structured content and off-site presence on Reddit, LinkedIn, and YouTube. Backlinks correlate at ~0 with AI citations. Read our full breakdown on why backlinks don't predict AI citations.
- Missing off-site presence. Not on Reddit, LinkedIn, or YouTube? Pick one and publish weekly for a quarter.
If this feels like a lot to run yourself, the AEO Audit does the baseline citation testing, the gap analysis, and the prioritized fix list for you.
Do I need to check citations for every AI engine?
Prioritize by where your buyers actually are. For B2B SaaS in 2026, the priority order is roughly:
- ChatGPT (with browse enabled) - most B2B research happens here
- Google AI Overviews - captures search-adjacent research
- Perplexity - power user favorite, disproportionately researchers and journalists
- Gemini - integrated with Google Workspace, growing enterprise share
- Copilot, Grok, Claude - smaller share but tracked by advanced teams
Start with the top 3. Add more engines as you scale your program.
What we track at Growth Pigeon
We run this monthly across 30 target prompts covering SaaS positioning, headline rewrites, ICP work, and AEO strategy. Current state (July 2026):
- Citation rate on target prompts: approximately 12% (up from 0% in Q1)
- Brand search volume: 4x quarter-over-quarter growth
- Perplexity citation coverage: strongest, roughly 20% of target prompts
- ChatGPT browse coverage: growing, roughly 10%
- Gemini + AI Overview coverage: catching up, roughly 8%
The AEO strategy that got us here is documented in the AEO for SaaS Founders pillar. Or run the same monitoring on your site with an AEO Audit.
FAQ
Is there a free tool to check AI citations?
No dedicated free tool as of mid-2026, but the manual method is free and takes 20 minutes for a first-pass baseline. Open ChatGPT, Perplexity, Gemini, and Google AI Overviews. Run your target prompts. Screenshot responses. Log the domains that get cited.
How accurate are AEO tracking tools?
Directionally accurate, not perfectly precise. AI responses are probabilistic, so tools that sample too infrequently or in only one geographic region can miss citations. Profound and Otterly.ai are the current category leaders and both have accuracy tradeoffs. Use them for trend tracking, not absolute counts.
Should I check citations for competitors too?
Yes. Competitor citation share is one of the most useful benchmarks in AEO. If competitor X is cited on 40% of your target prompts and you are cited on 5%, that is a specific gap to close. Look at what competitor X does on-page and off-site, and reverse engineer.
Does anonymous browsing affect citation testing?
Yes. ChatGPT personalizes based on account history. To get a citation baseline that mirrors an average user, test from a fresh incognito session, logged out, no history. That is closer to what a first-time user sees.
Can I game citations by paying for AI ad placements?
Not yet. There is no paid citations program. Some engines are testing sponsored responses, but sponsored is not the same as cited. Cited requires the model to select you as a trusted source, and no engine currently offers to sell that placement.
How do I set up citation tracking without a tool?
Google Sheet with columns: prompt, engine, date, cited (yes/no), notes. Run monthly. Track for 6 months to see meaningful trend data. This is the simplest workflow that produces real insight without any tool budget.
What's the difference between a citation and a mention?
A citation is when an AI response links to your URL as a source. A mention is when your brand name appears in the response text without a source link. Both are valuable. Mentions build brand awareness, citations drive direct traffic. Track both separately.
What to do this week
Three moves:
- List 10 buyer-intent prompts your customers actually ask AI. Not keywords. Real questions.
- Run each through ChatGPT (with browse), Perplexity, Gemini, and Google AI Overviews. Screenshot who gets cited.
- Pick the top 3 prompts you want to win. Note what the currently-cited sources have that you do not.
If you want a running dashboard instead of a manual spreadsheet, use Profound or Otterly.ai. If you want someone else to set up citation tracking and hand you the fix list, that is exactly what the AEO Audit does.