Self-serve global software
Automated API measurement of named global models. Directional evidence is labeled by surface fidelity and never presented as a signed-in consumer-client result.
See software pricing →Research-driven AI measurement
The China AI measurement firm
China AI measurement and evidence, captured on real consumer surfaces
We help international brands and agencies measure what Chinese AI assistants retrieve, cite, and recommend. Our self-serve software remains available for directional measurement across global API surfaces.
Run the free global diagnosis
1 free diagnosis (Quick) per account · No credit card · Website optional · Public portal
What we measure
Two measurement corridors, kept explicitly separate.
Managed China measurement
Evidence-grade measurement on a real, logged-in Doubao consumer client, with frozen questions and per-question evidence packets.
Explore China services →Watch GeoCheckTool turn a finding into action
Watch one GeoCheckTool run move from evidence to action.
Five guided states show what the product does. Every label and finding remains readable HTML; motion only advances the focus.
GeoCheckTool guided diagnosis
Interactive product preview
Business truth profile
Confirm facts before measuring answers.
Pricing, services, identity, locations, hours, credentials, and action paths remain customer-confirmed facts with provenance.
Confirmed business name
ClearFlow Plumbing
Customer-confirmedConfirmed pricing
$89 diagnostic visit
Customer-confirmedConfirmed services
water-heater repair · drain cleaning
Customer-confirmedConfirmed buyer question bank
Keep the full question bank visible.
01
What does a diagnostic plumbing visit cost, and which company is the best value?
2 analysis samples02
Which Denver plumber can handle an urgent leak and let me book the next step?
1 analysis samples03
Which residential plumber has trustworthy proof for water-heater repair?
2 analysis samples04
Does ClearFlow Plumbing serve Lakewood as well as Denver?
Next batch · not deleted05
How does ClearFlow Plumbing compare with RootHero for drain cleaning?
Next batch · not deletedBuyer question
Which Denver plumber can handle an urgent leak and let me book the next step?
Perplexity Sonar via OpenRouter / RAG API · Sample 1
“RootHero is the recommended 24/7 option. ClearFlow Plumbing is listed, but the answer gives no booking step.”
accepted · high
Accepted finding
The answer directs the urgent buyer to RootHero instead of ClearFlow Plumbing.
Buyer question
Which Denver plumber can handle an urgent leak and let me book the next step?
Accepted evidence
The answer directs the urgent buyer to RootHero instead of ClearFlow Plumbing.
Ready actions
Website fix
ready for review
Claim that ClearFlow Plumbing does not serve Denver contradicts confirmed fact.
Service area: Denver.
Place in: New service-area page, opening summary
Source task
ready for review
Add confirmed services to the customer-maintained business profile.
Confirmed services: water-heater repair, drain cleaning.
Place in: Business description and service fields
Independent action verification
Customer receipt and system evidence stay separate.
Customer report
Customer reported published
System verification
verified live match
https://clearflow.example/service-areas
A read-only check matched a frozen content anchor. This verifies the public change, not an AI outcome.
1
Observed improvement
1
No observed change
1
Observed regression
1
Not comparable
These are observed distributions from repeated API samples, not a stable ranking, market share estimate, or statistical significance claim.
A different job from a visibility dashboard
From confirmed truth to a verified retest—with the evidence chain intact.
GeoCheckTool does not stop at showing that a brand appeared. It preserves why a finding was accepted, turns it into controlled work, checks what actually went public, and retests the same frozen scope.
Confirm truth before scoring visibility
The customer confirms business facts first, so an answer can be checked for factual support—not only brand presence.
Keep every finding tied to its sample
Buyer question, exact answer span, citations, provider state, confirmed facts, and retry lineage remain connected.
Turn diagnosis into controlled work
Accepted findings become versioned website-repair and source-building tasks without publishing on the customer's behalf or taking over the CMS.
Verify first, then retest the same scope
Published, crawlable, indexed, mentioned, cited, and converted remain separate states; later runs keep the original method comparable.
Diagnose the question behind the query
Start with the surface or evidence gap you actually need to understand.
Each guide answers one search problem with the current GeoCheckTool method, then connects to the same official Quick Diagnosis workflow. No duplicate account, wallet, or second tool path is introduced.
AI search visibility diagnosis
Start from confirmed business facts and a complete buyer-question bank, then preserve repeated observations and evidence-linked actions.
Open guide →AI search monitoring
Track frozen buyer questions, repeated answers, mentions, citations, factual support, coverage, and comparable retests without inventing a stable rank.
Open guide →How to measure AI search visibility
Build a bounded baseline from frozen questions, named routes, repeated samples, and separate mention, citation, and lifecycle evidence.
Open guide →Google AI Overview checker
Observe the search-result surface through SerpApi and keep a legitimate no-surface result separate from failure.
Open guide →AI Overview SEO checklist
Use current Google eligibility guidance, keep generative-AI impressions separate from answer evidence, and review implementation without inventing a score.
Open guide →Perplexity visibility checker
Separate Sonar RAG/API answer text, cited URLs, fact support, and consumer-interface claims.
Open guide →Why is my brand missing from AI search?
Check question fit, entity clarity, facts, sources, access mode, and freshness before producing more content.
Open guide →AI mentions vs citations
Understand which evidence proves a mention, citation, factual support, or recommendation—and which evidence does not.
Open guide →AI visibility tools compared by workflow
Compare GeoCheckTool, Profound, AthenaHQ, and Peec AI by buyer stage, evidence model, action path, monitoring scope, and verified boundaries.
Open guide →Method and promises
Questions worth answering before you run GeoCheckTool.
What does GeoCheckTool need to start a diagnosis?
A real business name and city or service area are required. A website is optional. After sign-in, the customer confirms the Business Truth Profile before GeoCheckTool generates or measures buyer questions.
Does GeoCheckTool limit the buyer-question library?
No. The complete editable question bank remains visible and versioned. Pilot capacity limits only one frozen run scope: Quick selects up to 4 questions by intent, while Formal selects up to 8; uncovered and disabled questions remain explicit.
What does Quick measure?
The current Quick pilot uses n=3 repeated samples per selected question and surface for Perplexity Sonar via OpenRouter / RAG API and Google AI Overview observed via SerpApi. An unavailable surface is shown as not run, and a legitimate Google no-surface observation is not treated as a provider failure.
What does Formal measure?
The current Formal pilot freezes up to 8 selected questions and uses n=5 repeated samples per available surface for the two Quick surfaces plus a web-grounded model via OpenRouter. Each batch reserves 200 credits. Partial, failed, cancelled, or not-run coverage releases the reservation during Beta.
What happens after the diagnosis?
Accepted, evidence-linked findings can produce a website repair package and source-building package. Customers can execute the work, ask their original developer, or request our assistance. GeoCheckTool does not automatically take over a CMS or publish externally.
Does GeoCheckTool guarantee indexing, recommendation, ranking, or improvement?
No. GeoCheckTool can verify public actions with read-only checks and can compare a frozen set of questions and methods in a later repeated-sample retest. Results are distributions with uncertainty, not a stable rank or causal guarantee.