The Old Visibility Model Is Breaking
For two decades, visibility meant rankings. If you showed up in the top three Google results, customers found you. That still matters. But it’s no longer enough.
When someone asks ChatGPT which plumber to call, they don’t scan ten blue links and make a choice. They read a paragraph. If your business isn’t in that paragraph, you don’t exist in that decision.
That’s not a future problem. It’s already happening.
An AI citation is when a platform includes your business, named and briefly described, in a generated answer. The AI simply chose you as part of its response. No crawl required. No ranking to climb.
The difference from a featured snippet is worth understanding. A featured snippet pulls text directly from your website and credits the source. An AI citation synthesizes from multiple sources and may return your name without ever sending a visitor to your site. You get the mention without the click. That’s why citation tracking gives you a more complete picture than traffic analytics alone.
Why Tracking One Platform Tells You Almost Nothing
ChatGPT, Claude, Gemini, and Perplexity draw on different training data, use different retrieval methods, and weight local signals differently. The result is that each platform builds its own picture of your business. A strong showing on ChatGPT says nothing about how Perplexity describes you.
Each platform is its own audience. A customer asking Perplexity for a local contractor and a customer asking ChatGPT for the same service may get completely different names.
The practical rule: if you’re only checking one platform, you’re missing at least three audiences that are making decisions right now.
Five Numbers Worth Measuring
A five-metric framework gives you the complete picture:
- Mention rate. How often does your business appear when a relevant prompt is run across all four platforms? A high rate on ChatGPT with near-zero on Perplexity points specifically to a platform gap.
- Sentiment. When you’re mentioned, how are you described? “Smith Plumbing is an option” and “Smith Plumbing is highly recommended” are both citations. Only one is working for you.
- Factual accuracy. Is the AI citing the right address, services, and credentials? Hallucinated details are common and actively harmful. A Hallucination Risk Score flags the specific claims platforms are getting wrong about your business.
- Share of voice. Which competitors appear alongside you? How often are you named relative to them across category and local queries?
- Local capture. When city-specific discovery prompts run for queries like “best [service] in [city],” do you appear? This is where most local businesses have the biggest gap, and where most enterprise AEO tools fall short entirely.
Manual Tracking Has a Real Cost
You can do this by hand. Open each platform, run a consistent set of prompts, record what appears. Budget 2 to 4 hours per platform per month if you want any meaningful trend data.
The math breaks down fast. Four platforms. Two prompt categories: branded and local discovery. Monthly cadence. That’s 8 to 16 hours of manual effort every month, with no consistent scoring, no trend charts, and no alerting when something changes. See how the leading AI citation tracking tools compare if you’re deciding whether to automate.
| Approach | Cadence | Scoring | Trend data | Cost |
|---|---|---|---|---|
| Manual | As often as you have time | None | None | 8–16 hrs/month |
| Essentials | Monthly automated | Scored 0–100 | Yes | $99/mo |
| Growth | Monthly automated | Scored 0–100 | Yes + share of voice | $199/mo |
Start With a Baseline
Before you can measure progress, you need a current, accurate picture of how each platform describes your business today.
The fastest way: run your free AI citation baseline. It tests all four platforms with structured prompts, scores each of the five dimensions above, includes a Local Capture Score for city-specific discovery queries, and emails you the full report within a few hours. No account required. No credit card.
Once you have that baseline, you can make changes: update directory listings, publish new content, correct outdated information. Then measure whether any of it moved the needle when you re-run the audit.
Without the baseline, you’re optimizing blind.
For more on the satellite questions around this topic, see how to appear in AI search results and how to get cited in AI search results.