Score each result on the same AI Visibility ladder PR teams use in modern reporting: Top, High, Medium, Low, or Not Yet Indexed -and fix the gaps those tiers reveal. A five-minute manual audit gets you a baseline; scheduled tracking turns it into a trend you can put in front of a client.
ChatGPT now has over 810 million weekly active users. Many are asking it about brands, products, and services. When someone asks "what's the best PR reporting tool?", ChatGPT generates an answer based on what it's found across the web - including your press coverage.
If your coverage appears in that answer, it's working. If it doesn't, you're invisible to a growing share of your audience.
And here's what makes it worse: Muck Rack's research found that the overlap between the journalists' PR teams pitch most and the ones AI actually cites is just 2%. The industry is still optimizing for an audience that has quietly moved.
Let's fix that.
Before you can check whether your coverage appears, you need to understand how ChatGPT picks its sources. It's not random.
When someone asks ChatGPT about your industry, the AI:
Earned media drives 84% of all AI citations across ChatGPT, Claude, and Gemini, making it the dominant factor in Generative Engine Optimization (GEO) for brand visibility in AI-generated answers.
You do not need a fake precision statistic to act. You need a repeatable way to see whether your placements and your brand are present for the prompts that matter.
Most people ask, “Am I visible in AI?” and expect one number. There isn’t one useful number. There’s a stack, and the layer you can see depends on the prompt type.
For any answer, the model draws on a set of cited sources: specific URLs, the domains behind them, and the type of outlet (trade press, news, company site, aggregator, forum). Before you ask whether your brand is winning, you need to know which outlets the model leans on in your category. The Sources layer is the map of where the conversation is actually being built.
Of those cited sources, how many are yours? This is the narrowest and most concrete metric: how often your own articles or domain show up as a citation behind an answer. It tells you whether your specific content is feeding the model, not just whether your name came up.
Across all the citations the model pulls, how often is your brand mentioned at all, by anyone, relative to competitors? This is broader than Share of Citations because it counts every source that references you, not only the ones you own. A trade article that lists five vendors and mentions you once contributes to your Share of Media even though it's not your article.
When a prompt asks for a recommendation, the model returns an ordered list of brands. Where you land in that list, and how consistently you appear across prompts, is your Ranking. Aggregate it and you get Share of Voice: your slice of the recommendation real estate.
But there’s a catch. The metric you can measure is determined by the kind of question you ask. Informational prompts ("what is AI visibility tracking?") produce Sources, Share of Citations, and Share of Media - but no Rankings. Comparison prompts ("best AI visibility tracking tools") produce Rankings and Share of Voice.
Same category, same brands, completely different measurement surface. Run the wrong prompt type for the metric you want and you'll draw the wrong conclusion.
Build 20-50 prompts your audience would ask. Mix prompt types to capture the full measurement stack:
Informational prompts (measure Sources, Share of Citations, Share of Media):
Comparison prompts (measure Rankings, Share of Voice):
For each prompt–platform pair, log:
Context shorthand still helps:
Do not invent a 0–5 “product score.” Score the way modern PR AI reporting does: assign an AI Visibility tier to what you observed.
These rows illustrate how to fill Sources Cited using domains that actually publish on these topics (SERP-sampled references, not a claim that ChatGPT cited them in your account yesterday). When you run the audit, replace the source column with whatever the model shows you that day.
PRCoverage was built specifically for this workflow. Here's how it works:
Client one-liners
Forget the tools for a second. You can answer "is my PR coverage showing up in ChatGPT?" in five minutes with nothing but a browser and a spreadsheet. Here's how.
Run these ten prompts in ChatGPT (GPT-5.5 recommended). Mix informational and comparison prompts to capture the full measurement surface - informational prompts reveal whether your content is being cited, while comparison prompts reveal whether you're ranking.
Informational prompts (test whether ChatGPT knows about your brand and cites your coverage):
Comparison prompts (test whether ChatGPT recommends you by name in ranked lists):
For each prompt, open ChatGPT, enter it exactly as written, wait for the full response, and score your brand on a 0-5 scale:
Step 1: Total your scores. Add up all ten scores. Maximum possible = 50 (10 prompts × 5 points each).
Step 2: Calculate your AVS.
AVS = (Your Total Score / 50) × 100
Step 3: Read your result.
Copy this into a spreadsheet and fill it in:
Here's the part most people miss: the distribution of your scores matters more than the total.
Scenario A: High scores on informational prompts, low on comparison prompts.
ChatGPT knows about you and cites your content - but when someone asks for a recommendation, you don't make the cut. This means your content is educational but not positioned as a solution. Fix: add comparison tables, "best for" framing, and decision-matrix content to your coverage.
Scenario B: Low scores on informational prompts, high on comparison prompts.
You're ranking in recommendation lists but ChatGPT doesn't cite your content as a source. This means other people's coverage is mentioning you, but your own content isn't authoritative enough to be cited. Fix: create extraction-friendly content with definition blocks, cited statistics, and structured data.
Scenario C: Low scores across the board.
ChatGPT doesn't know you exist. This is a content volume and authority problem. You need more coverage on high-DA sites, not better tracking. Fix: pitch journalists, publish on authoritative platforms, build backlinks.
Scenario D: High scores across the board.
You're winning. But check which prompts you score 5 on vs. 3 - the gap between "top 3" and "#1 with positive framing" is the difference between being mentioned and being the default recommendation. Fix: create content specifically targeting the prompts where you score 3-4 to push them to 5.
If you scored below 15: You don't have a tracking problem. You have a content problem. Jump to "What If My Coverage Isn't Showing Up?" and work through the reasons.
If you scored 15-55: You're in the zone where small improvements compound fast. Set up PRCoverage.AI ($50/mo) to track your score weekly and see which changes move the needle. The manual audit gives you a snapshot; automated tracking shows you the trend.
If you scored above 55: You're ahead of most competitors. But AI models update constantly - what works today might not work next month. Schedule monthly re-audits to catch shifts before they become problems.
The key insight: This audit takes five minutes. Doing it once tells you where you stand. Doing it monthly tells you whether you're improving. Doing it weekly (with automation) tells you what's working in real time. Start with the five-minute version. Graduate to automation when the manual process starts feeling tedious.
If you're a PR team or freelancer: Start with PRCoverage. It's the only tool built for your workflow - import coverage, see which pieces influence AI answers, generate client reports.
If you're an agency managing multiple clients: OtterlyAI or SE Ranking. Both offer multi-client workspace management and API access.
If you need social listening + AI visibility: Brand24. It's expensive, but it's the only tool that combines 25M+ social source monitoring with 9-platform AI tracking.
If you're enterprise: Conductor. Full AEO suite with AgentStack, LLM apps, and enterprise security.
If you're on a budget: ZipTie at $69/mo gets you started with the three major AI platforms.
AI models extract content in chunks. If your coverage is walls of text without structure, AI can't parse it.
Fix: Create extraction-friendly content - short paragraphs, definition blocks, comparison tables.
AI models prefer citing authoritative sources. If your coverage is on low-DA sites, AI may ignore it.
Fix: Earn coverage from high-DA publications (PRWeek, Forbes, TechCrunch).
If your site blocks GPTBot, AI can't access your content.
Fix: Check robots.txt and ensure GPTBot, ClaudeBot, and Google-Extended aren't blocked.
AI can't cite coverage that doesn't exist.
Fix: Pitch journalists, publish industry insights, build media relationships.
If competitors have stronger AI visibility, they'll be cited instead.
Fix: Create more content, build more authority, track and improve.
You ran informational prompts and saw no rankings - but rankings only appear on comparison prompts. Your brand might have a strong Share of Citations and simply never asked the model to rank anyone.
Fix: Mix informational and comparison prompts in your tracking. Both are real - they measure different surfaces.
AI models favor fresh, frequently updated content. Publish weekly if possible.
Use AI visibility tracking to measure what works and iterate.
Ask ChatGPT prompts related to your industry and brand. If your brand appears in the response, your coverage is showing up. For systematic tracking, use PRCoverage ($50/mo).
Common reasons: content isn't extractable, low source authority, blocked by robots.txt, no coverage exists, competitors dominate, or you're measuring the wrong prompt type. Fix by creating extraction-friendly content, earning high-DA coverage, and optimizing for AI crawlers.
Create extraction-friendly content (short paragraphs, definition blocks, tables), build source authority (earn high-DA coverage), optimize for AI crawlers, publish consistently, and track what works.
Yes. Create a spreadsheet of 20-50 prompts, run them monthly, record results. Manual process takes 3-5 hours/month. For automation, use PRCoverage.
Monthly minimum. Weekly for competitive markets. AI models update frequently.
A mention is when ChatGPT includes your brand name in its response. A citation is when ChatGPT includes a link to your content as a source. Citations are more valuable because they drive traffic and indicate authority.
It depends. ChatGPT's training data has a cutoff date, so new coverage may not appear immediately. However, ChatGPT also searches the web in real-time for some queries, so recent coverage can appear quickly. Track monthly to see trends.
Traditional tools monitor press mentions and journalist contacts. PRCoverage focuses specifically on whether your coverage appears in AI-generated answers - a different layer of visibility that traditional tools don't measure.
PRCoverage tracks ChatGPT and Gemini today, with Claude support coming soon. Manual tracking works for any platform but takes significantly more time. Most PR teams start with ChatGPT since it has the largest user base.
Share of Media shows clients exactly what percentage of AI conversations mention your brand vs. competitors. This is concrete, quantifiable proof that your PR work drives results - something traditional coverage reports can't demonstrate.
They serve different purposes. PRCoverage is complementary to traditional monitoring. Use your existing tool for media alerts and journalist outreach. Add PRCoverage to measure whether that coverage is actually visible in AI answers.
This guide was last updated June 2026. PRCoverage tracks your brand's visibility across ChatGPT, Gemini, and Claude - automatically. From $50/month. Start your free trial →
To check if your PR coverage is showing up in ChatGPT, run prompts your buyers would actually ask, then record two things for every answer: whether your brand appears, and which sources the model cites.