All Posts
How-To April 27, 2026 31 views Ninar AI

How to Measure Your Brand's Visibility in ChatGPT, Gemini, Claude, and Perplexity

The exact step-by-step methodology for measuring how AI engines see your brand. Prompts to test, scoring framework, common mistakes, and how to automate the whole thing.

If you only have 30 seconds: Measuring AI visibility takes four steps: pick a probe set of buyer-stage queries, run them across all 10 AI engines, score how often your brand appears as a named recommendation, and tag every Ghost Intent. Manual measurement works for one-time spot checks. For continuous measurement, Ninar AI automates the entire flow with a free tier you can use without a credit card.

Why Measurement Matters

You can't improve what you don't measure. Every AI visibility program starts with a baseline: where does your brand stand right now across the engines that matter? Without that baseline, every subsequent action is a guess.

The challenge is that AI visibility doesn't measure like SEO. There's no rank position to track. There's no clear “impression count.” Buyers ask AI engines questions and the AI returns named recommendations — either you're in the answer or you aren't. Measurement requires a structured approach.

This guide walks through the exact methodology, with the specific prompts, scoring framework, and tools you can use today.

Step 1: Build Your Probe Set

A probe set is the list of queries you'll test. The right probe set covers the entire buyer journey for your category, using language that real buyers use.

Cover All Eight Buyer-Journey Intents

Build at least one probe per intent:

Add Multiple Variations Per Intent

Buyers ask the same question many ways. For each intent, run 2-3 phrasing variations to get a more reliable signal. The variations matter because AI engines often interpret subtly different phrasings differently.

Use Real Buyer Language

Skip industry jargon. Probe sets should sound like a buyer typing into ChatGPT, not like marketing copy. “Best AI visibility platform for a small marketing team” is a real query. “Optimal Generative Engine Optimization solution for SMB marketing teams” is not.

Step 2: Run Probes Across All 10 AI Engines

This is where manual measurement gets painful. The 10 engines that matter in 2026:

For a probe set of 8 intents with 3 variations each (24 probes), running all 10 engines manually is 240 individual queries. Most teams give up before finishing.

Ninar AI runs all 240 probes concurrently in a single scan and writes the responses to its database for analysis — the same workflow that would take a person two days takes the platform a few minutes.

Step 3: Score Each Probe

For every probe response, capture three things:

Was Your Brand Mentioned?

The most basic signal. Did the AI engine name your brand at all? If yes, mark a 1. If no, mark a 0.

How Was It Mentioned?

Distinguish three mention types because they're worth very different amounts:

What Sources Were Cited?

For engines that show citations (Perplexity, AI Overviews, Gemini), capture which domains the AI cited when answering. This becomes critical for diagnosis — the same domains that show up for competitor recommendations are the ones you need to land in.

Step 4: Calculate Your Visibility Score

A reasonable formula for a single-engine visibility score (0-100):

Sum across all probes for a given engine, normalize against the maximum possible score, multiply by 100. Repeat per engine. Aggregate by averaging across engines for a unified score.

For more advanced scoring, weight Decision-stage probes (Pricing, Recommendation, Comparison, Top Tools) higher than Awareness-stage probes (How-To, Trust). Ninar AI uses an intent-weighted scoring model with Decision intents at 1.2x and Awareness intents at 0.85x weight.

Step 5: Tag Every Ghost Intent

For each intent category, look at the per-intent score. If a specific intent scores zero or near-zero across multiple engines but your brand appears in awareness queries, that's a Ghost Intent — AI knows you exist but won't recommend you for that specific buyer query.

Ghost Intents are the most actionable diagnostic output. They tell you exactly which content gaps to fill first.

Step 6: Run Variance Tests

AI responses are stochastic. The same probe run twice can return different answers. To make sure your scores are reliable, run variance tests: run the same probe 3-5 times against the same engine and check whether the brand mention pattern is stable.

Three stability bands:

Ninar AI runs variance tests automatically and surfaces the stability band per probe.

Step 7: Compare Against Competitors

Visibility scores are most useful in relative context. Run the same probe set with your competitors' brand names. If the same Decision-stage prompts surface your competitor 80% of the time and your brand 5%, you have specific evidence of where you're losing.

Manual competitor benchmarking is tedious. Ninar AI auto-detects competitors mentioned in your probes and scores them in parallel, giving you a competitive map without separate scans.

Step 8: Track Over Time

One scan is a baseline. Real measurement is the trend. Set up regular scans — weekly for active programs, monthly for maintenance — and track:

Sentiment Velocity is particularly valuable as a leading indicator. A brand whose sentiment is decelerating in AI answers will see its visibility score decline 30-60 days later. Catching the deceleration early lets you intervene before scores drop.

Common Measurement Mistakes

Mistake 1: Scanning Only One or Two Engines

Single-engine scores give a falsely confident picture. The same brand can score 78 on ChatGPT and 23 on Perplexity. Always measure across all 10 engines.

Mistake 2: Using Brand-Specific Probes Only

Asking AI “Tell me about [your brand]” tests awareness, not recommendation. The probes that matter are category-level queries that don't name your brand — those reveal whether AI surfaces you spontaneously.

Mistake 3: Running Probes Once Without Variance Testing

AI responses are stochastic. A single probe run isn't statistically reliable. Always run multiple iterations to establish a stability band.

Mistake 4: Ignoring Buyer-Journey Intent

A high aggregate score can hide critical Ghost Intents. A brand can win Awareness queries (How-To, Trust) and lose every Decision query (Pricing, Recommendation, Comparison). Per-intent measurement is non-negotiable.

Mistake 5: Treating Visibility as a Snapshot, Not a Trend

One scan tells you where you are today. Three scans over three months tell you whether you're improving. The trend matters more than the baseline.

How to Automate the Entire Workflow

Manual measurement at the depth this guide describes — 10 engines, 8 intents, 3 variations, 5 variance runs, weekly cadence — is unsustainable for any team. Ninar AI automates the entire workflow:

The free tier covers 2 engines and is enough for a baseline measurement. The Pro tier ($79/month) covers 4 engines and includes the diagnostic suite. The Scale ($299) and Enterprise ($599) tiers cover all 10 engines.

Frequently Asked Questions

How long does a manual AI visibility measurement take?

For a probe set of 24 probes (8 intents x 3 variations) across 10 engines, manual measurement takes 1-2 days of focused work, with another day for analysis. For weekly cadence, manual measurement is impractical for any team. Ninar AI runs the same workflow in minutes.

Can I use ChatGPT to measure my own AI visibility?

You can run individual probes against ChatGPT to spot-check, but you can't get a complete picture from a single engine. AI visibility requires multi-engine measurement because each engine weights signals differently.

What's the simplest measurement to start with?

Run 5 category-level probes (no brand name in the query) across ChatGPT and Gemini. Note whether your brand appears in the answers. This 10-minute exercise gives you enough signal to know if you have an AI visibility problem worth investigating further.

How do I know if my measurement is statistically reliable?

Run each probe 3-5 times. If your brand appears in most runs, the signal is reliable. If it appears in only 1-2 runs out of 5, your visibility is volatile and the score is less trustworthy. Ninar AI automates this with the Anchored / Established / Emerging stability bands.

Should I measure across multiple geographies?

Yes if you operate in multiple markets. AI engines respond differently in different geographies because grounding signals (citations, search results, regional sources) differ by region. Run city-specific probes for every market you serve. Ninar AI's local module handles 500+ cities natively.

How often should I rescan?

For active AI visibility programs: weekly. For maintenance mode: monthly. For rapid-iteration content programs: daily. Ninar AI offers all three frequencies.

Run your first measurement in 60 seconds. Ninar AI's free tier gives you a baseline AI visibility scan across two engines — no credit card required. Start your free scan →
measure ai visibility how to chatgpt gemini perplexity claude ninar ai

© 2026 Ninar AI — Your brand's voice inside every AI engine.