Visibility
How often your brand appears in neutral discovery answers.
AI visibility benchmark · Four web-grounded models
Find where AI models overlook your brand—and what to fix next.
A $9 AI visibility benchmark across OpenAI, Claude, Gemini, and Grok: missed buyer questions, competitors shown instead, sources models trust, and five evidence-backed actions—no subscription.
Private by default · 30-day answer retention
A real deliverable, before you buy
Metrics, missed questions, sources, and five next actions.
Example benchmark
Northstar analytics
Discovery visibility
38%
91% coverage
Owned citations
24%
73 eligible
Model visibility0–100%
OpenAI
46%
Claude
34%
Gemini
31%
Grok
39%
Every metric and recommendation links back to stored evidence.
OpenAI
GPT-5.4 mini · Web-grounded search
Claude
Claude Sonnet 5 · Web-grounded search
Gemini
Gemini 3.1 Flash Lite · Web-grounded search
Grok
Grok 4.5 · Web-grounded search
Featured on
Find 100 Questions around the web.
These launch, product, and search-performance profiles provide independent places to verify the product. Inclusion is not a rating, customer review, or endorsement.
Original research
25 products, 80 grounded answers, and 210 cited domains
The 2026 AI Visibility Index publishes its frozen questions, provider splits, source frequency, answer-level evidence, protocol, adjudications, and file hashes for independent review.
Know what you receive
A client-ready audit, not another monitoring dashboard.
Explore the AI visibility audit25 buyer questions
Four web-grounded providers
Competitor comparison
Answer and citation evidence
Five prioritized actions
PDF and CSV exports
The benchmark
A defensible answer to “are we visible?”
The shared test uses 20 neutral discovery questions that avoid planting your name in the prompt, plus 5 diagnostic questions that ask directly. Each model receives the identical 25-question set, making the 100 planned provider answers a directional comparison.
See definitions, eligibility, and limitationsProminence
Whether you lead the answer, make a shortlist, or appear incidentally.
Share of voice
Your answer-level mentions compared with selected competitors.
Citations
How often eligible answers cite your claimed domain.
Grounding required
A response without valid web sources is excluded from scoring and remains visible in coverage.
Evidence becomes action
See the buyer questions you missed, who appeared instead, which sources recur, and what page or third-party mention is likely needed.
Same frozen test
All four models receive the same 25 questions, locale, and prompt version; the run freezes its timestamp with the results.
Guides and resources
Start with the free checker or the full guide library
The benchmark is the paid product. These guides explain the measurement model when you need more context before buying.
Choose the right measurement model
Best for a controlled benchmark—not an always-on dashboard.
100 Questions trades monitoring frequency for a frozen test, inspectable evidence, and a bounded project cost. That makes it a strong fit for specific decisions, but not every AI visibility job.
Use 100 Questions for
- A point-in-time baseline before an AI visibility project
- A client-ready audit with inspectable answer and citation evidence
- A like-for-like rerun after meaningful technical, content, or authority work
Choose continuous monitoring when
- It is not continuous monitoring and does not send daily alerts.
- It does not currently test Perplexity or reproduce consumer chat interfaces.
- It is a directional 25-question sample, not a statistically representative market ranking.
Product questions
Clear answers before you buy.
Who is 100 Questions best for?
100 Questions is best for consultants, agencies, and in-house teams that need a bounded AI visibility baseline, an evidence-linked client deliverable, or a comparable rerun after implementation work.
When is continuous AI visibility monitoring a better choice?
Choose continuous monitoring when you need daily or weekly trend lines, alerts, a large configurable prompt program, Perplexity coverage, or ongoing multi-project reporting. 100 Questions is deliberately a point-in-time benchmark rather than an always-on dashboard.
How much does a 100 Questions benchmark cost?
The first benchmark costs $9. After that, one benchmark costs $15, three cost $39, and ten cost $99. There is no subscription, credits remain valid for 12 months, and Stripe shows applicable taxes before payment.
Does 100 Questions guarantee placement in AI answers?
No. It measures a time-stamped, API-grounded sample and preserves the evidence behind it. AI answers vary by provider, model, search results, prompt, location, and time, so no credible tool can guarantee future mentions or citations.
Simple, prepaid pricing
One credit. One complete benchmark.
No subscription, seat fees, or provider add-ons. Every run includes the audit, evidence, prioritized action plan, and exports.
First benchmark
$9
A complete first benchmark at the introductory first-purchase price.
1 complete benchmark
Three benchmarks
Best value$39
Baseline, implementation check, and final measurement.
3 complete benchmarks
Choose packageTen benchmarks
$99
Flexible capacity for consultants and agencies.
10 complete benchmarks
Choose packageIntroductory price is limited to the first purchase. Credits are valid for 12 months. Normal single-benchmark price after the introductory purchase is $15. Taxes may apply.
Prepaid, not a subscription
Buy one run. Review 30 days of evidence.
Packages start at $9. Stripe confirms the price and applicable taxes before payment.
Need details first? Read the FAQ.