AI Search Visibility
Observed mentions, citations, or presence across a defined prompt and answer-surface sample.
Visibility starts with a recorded observation: one system included—or omitted—a brand, source, link, or product. Remove conditions or misses and the percentage is not auditable.
Readiness is separate: a crawl tests storefront conditions; a defined output panel measures presence. Search Console, analytics, and conversions add different evidence, and none substitutes for another.
What counts as AI search visibility?
AI search visibility is a measured observation of whether a defined AI surface mentions a brand, cites or links a domain, or presents a product for specified prompts and conditions. The result belongs only to the sampled systems, prompts, dates, locales, devices, and account states; it is not universal coverage.
ChatGPT Search and Google AI features are distinct surfaces. The AI citation glossary distinguishes a linked source from a mention. Engines vary; one run is one observation, not coverage.
How is visibility different from readiness and business outcomes?
Visibility, readiness, and business impact answer different questions. Visibility asks what appeared in sampled outputs; readiness asks whether public storefront conditions are implemented; business measurement asks what users did afterward. A readiness score cannot substitute for observed citations, and an observed citation cannot prove incremental traffic, conversion, or revenue.
| Measurement layer | Evidence | Valid statement | Invalid shortcut |
|---|---|---|---|
| Observed AI visibility | Fixed output panel including misses | Presence rates for that panel | Universal rank or market share |
| Storefront readiness | Deterministic public audit | Conditions found at scan time | Actual visibility or citation likelihood |
| Search performance | Search Console metrics | Documented Google Search activity | Complete prompt coverage |
| Business outcomes | Instrumented visits, events, or orders | Observed owned outcomes | Causal effect of a citation |
Search Essentials covers Search eligibility, not prompt visibility. StoreCited’s AI Visibility Score is readiness, not observed presence.
Which metric definitions make results auditable?
Every visibility metric needs a named unit, numerator, denominator, and miss rule before collection begins. Count each prompt-system-condition observation once, preserve failed or absent outputs, and define whether one answer can contain multiple mentions or links. Changing the denominator after seeing results turns a measurement into a selected anecdote.
| Metric | Numerator | Denominator |
|---|---|---|
| Mention rate | Eligible observations naming the brand | All planned eligible observations, including misses |
| Citation rate | Eligible observations citing the store | All planned eligible observations, including misses |
| Linked-domain share | Links to the store domain | All observed source links in the panel |
| Answer accuracy | Verified factual claims | All factual claims assessed with the stated rubric |
| Product-card presence | Eligible shopping observations with a product card | All planned eligible shopping observations, including misses |
Predefine output-level versus item-level counting. Report raw numerators and denominators; unlike units are not comparable.
How should an ecommerce prompt panel be designed?
A useful prompt panel is a fixed sampling protocol, not a handful of favorite queries. Freeze the system, surface, prompt text, product or category intent, date, locale, language, device, and account condition; then run all planned observations and record misses. Repeating the same design reveals change within that panel only.
Use the measurement guide and AI rank tracking for sampling rules:
- Name systems, surfaces, and displayed labels.
- Freeze prompts and buyer intents.
- Fix locale, language, device, and account condition.
- Set dates, cadence, and comparison windows.
- Run all planned observations consistently.
- Save outputs, screenshots, links, and product cards.
- Separate misses from technical failures.
- Calculate only the predefined metrics.
OpenAI bots, Perplexity bots, and the Robots Exclusion Protocol govern access, not prompt presence. A crawler pass cannot replace the panel.
How do Search Console, analytics, and conversions fit?
Search Console, referral analytics, and conversions are separate evidence streams that complement prompt observations. Search Console reports Google Search activity under its own definitions; analytics records instrumented site behavior; conversion systems record business events. None reconstructs every AI answer, and prompt-panel presence does not itself establish a click or sale.
Use the Search Analytics API and Performance report within their documented scopes; use Google Analytics for instrumented site activity. Keep referral, conversion, and prompt datasets separate without a documented join.
Compare like dates and segments without claiming causation; demand, seasonality, interface changes, privacy limits, attribution, and releases can move datasets independently.
How should outputs and claims be verified?
Verification checks both the captured output and any claim you plan to report. Save the complete response, visible links or cards, screenshot, timestamp, and conditions; open cited sources; compare product facts with canonical pages; and label unverifiable statements. Schema can help describe entities, but its presence does not prove or cause citation.
Article, Product, and JSON-LD describe representations, not citation causality. Follow FTC marketing guidance by retaining evidence and stating material limits.
Verification must preserve negative evidence too. Save outputs without the brand, outputs with inaccurate claims, broken citations, absent product cards, and technical failures under the predefined treatment rule.
What belongs in an AI visibility report?
A defensible report publishes the panel definition before the headline rate. Include systems and surfaces, prompt list, dates, locales, account conditions, metric formulas, raw counts, misses, source-verification method, accuracy rubric, known exclusions, and comparison windows. Show readiness, Search Console, referrals, and conversions in separate sections with their own scopes.
Reporting checklist:
- Panel version, owner, and evidence dates
- Systems, surfaces, prompts, and buyer intents
- Locale, language, device, and account conditions
- Metric formulas with raw numerators and denominators
- Misses, technical failures, and treatment rules
- Screenshots, links, product cards, and evidence archive
- Citation verification and answer-accuracy rubric
- Separate readiness, Search Console, referral, and conversion sections
- Limitations, changes, and alternative explanations
Publish enough method detail to reproduce the panel without exposing credentials or customer data. StoreCited’s research disclosures illustrate bounded source and sample notes; they are not a substitute for your prompt-panel evidence.
What can StoreCited measure?
StoreCited is an Organization providing a deterministic, point-in-time audit of observable public Shopify or DTC storefront readiness. Its proprietary AI Visibility Score is a readiness composite only—not live visibility, citation likelihood, market share, rank, or prediction. The scan does not observe prompts, citations, selected competitors, traffic, conversions, or revenue.
StoreCited cannot monitor live prompts or citations, query proprietary indexes, identify actual selected competitors, or guarantee outcomes. Run the free StoreCited readiness scan for a dated readiness checklist, then measure visibility with a separate defined panel and preserve the boundary in every report.