SEO/GEO visibility metrics

Metrics

• geo - instrumented

  AI citation share

  ai-citation-share

  Answer engines can satisfy informational queries without a click; citation share measures whether Visa Atlas is becoming the named source.

  Target

  At least 40% of fixed benchmark answers cite visaatlas.org.

  Cadence

  Monthly, plus after major data/report launches.

  Evidence surfaces

  • /citation-benchmark
  • /api/public/citation-benchmark
  • /answer-capsules
  • /api/public/answer-capsules
  • /research/canada-express-entry-category-tracker
  • /api/public/express-entry-categories
  • /ai-agents

  Commands

  • npm run kpi:citation-share
  • npm run kpi:citation-share:perplexity
  • npm run kpi:citation-accuracy

  Caveats

  • Runs offline with user-provided API keys; the website never calls an LLM at runtime.
  • Citation detection looks for the configured brand domain in structured citations or answer text.
  • Accuracy grading is conservative answer-key triage; partial or failed rows require human review before publication.

• geo - instrumented

  AI/search crawler access

  crawler-access

  GA4 cannot see crawler fetches. Server logs are the only honest way to separate crawler access from referral traffic.

  Target

  Known search and AI crawler user agents are allowed and visible in server logs when they crawl.

  Cadence

  Monthly and after robots, middleware, CDN or hosting changes.

  Evidence surfaces

  • /robots.txt
  • /research/crawler-ai-retrieval-access-audit-2026
  • /api/public/crawler-access
  • /ai-agents
  • /llms.txt
  • /llms-full.txt

  Commands

  • npm run audit:botlog
  • npm run seo:botlog -- logs.txt
  • npm run seo:botlog -- logs.txt --json
  • npm run audit:visibility-metrics

  Caveats

  • Gemini training crawls can appear as Googlebot; Google-Extended is a robots token, not a user-agent.
  • User-agent strings can be spoofed, so log classification is directional rather than forensic proof.
  • The bot-log parser is regression-tested against the public crawler-access policy, but live counts require user-supplied server logs.

• trust - instrumented

  Source freshness

  source-freshness

  Freshness is the core wedge against stale immigration blogs and model training cutoffs.

  Target

  Zero stale or invalid dated source records in the public freshness report.

  Cadence

  Every build and before publishing major research or policy-change assets.

  Evidence surfaces

  • /freshness
  • /api/public/freshness
  • /methodology

  Commands

  • npm run audit:stale
  • npm run audit:freshness

  Caveats

  • A stale record does not prove the government fact is wrong; it means the Visa Atlas review date exceeded the public SLA.
  • High-risk legal and fee changes still require source verification before publication.

• geo - instrumented

  Open-data discoverability

  open-data-discoverability

  The site wins by being the reusable source layer that journalists, researchers and answer engines can cite.

  Target

  Every public data asset is listed in /data, /api/public, DataCatalog/Dataset JSON-LD, API Link headers and LLM discovery files.

  Cadence

  Every build and whenever a public endpoint is added.

  Evidence surfaces

  • /data
  • /api/public
  • /api
  • /api/openapi.json
  • /use-our-data
  • /research/topical-authority-internal-link-map-2026
  • /api/public/topical-authority-map
  • /research/seo-geo-audit-evidence-pack-2026
  • /api/public/seo-geo-audit-evidence
  • /research/seo-geo-execution-readiness-scorecard-2026
  • /api/public/execution-readiness
  • /research/search-demand-action-map-2026
  • /api/public/search-demand-actions
  • /research/crawler-ai-retrieval-access-audit-2026
  • /api/public/crawler-access
  • /answer-capsules
  • /api/public/answer-capsules
  • /research/publisher-citation-kits-2026
  • /api/public/publisher-citation-kits
  • /research/authority-distribution-plan-2026
  • /api/public/authority-distribution-plan
  • /research/external-data-repository-mirrors-2026
  • /api/public/external-repository-mirrors
  • /research/knowledge-graph-entity-seeding-2026
  • /api/public/knowledge-graph-entities
  • /research/expert-reviewer-entity-program-2026
  • /api/public/expert-reviewer-entities
  • /research/reviewedby-schema-rollout-2026
  • /api/public/reviewedby-schema-rollout
  • /research/high-risk-page-review-acquisition-2026
  • /api/public/high-risk-page-review-acquisition
  • /research/p0-reviewed-page-refresh-2026
  • /api/public/p0-reviewed-page-refresh

  Commands

  • npm run audit:data-front-door
  • npm run audit:openapi
  • npm run audit:topical-authority-map
  • npm run audit:seo-geo-audit-evidence
  • npm run audit:execution-readiness
  • npm run audit:search-demand-actions
  • npm run audit:data-citation
  • npm run audit:answer-capsules
  • npm run audit:authority-distribution
  • npm run seo:authority-distribution -- authority-distribution-log.jsonl --json
  • npm run seo:publisher-citation-kits -- outreach-log.jsonl --json
  • npm run export:repository-mirrors
  • npm run seo:repository-mirrors -- repository-mirror-log.jsonl --json
  • npm run seo:knowledge-graph -- kg-entity-log.jsonl --json
  • npm run export:knowledge-graph -- kg-entity-log.jsonl
  • npm run seo:expert-reviewers -- reviewer-evidence-log.jsonl --json
  • npm run export:reviewedby-schema -- reviewer-evidence-log.jsonl
  • npm run seo:review-acquisition -- reviewer-acquisition-log.jsonl --json
  • npm run seo:p0-page-refresh -- p0-reviewed-page-refresh-log.jsonl --json

  Caveats

  • Open data helps competitors only if they can match freshness and provenance; attribution turns reuse into entity signals.
  • A dataset is included only when its source chain and reuse terms are visible.

• technical - instrumented

  Rendered SEO/GEO surface

  rendered-seo-surface

  Source-template checks are not enough; crawlers and AI retrievers consume the static HTML and JSON produced by the build.

  Target

  Representative built pages expose canonical URLs, Article/Dataset schema, citations, source dates and no misleading ClaimReview.

  Cadence

  Every production build.

  Evidence surfaces

  • /api/public
  • /data
  • /research/technical-seo-control-plane-2026
  • /api/public/technical-seo-controls
  • /research/topical-authority-internal-link-map-2026
  • /api/public/topical-authority-map
  • /research/seo-geo-audit-evidence-pack-2026
  • /api/public/seo-geo-audit-evidence
  • /research/seo-geo-execution-readiness-scorecard-2026
  • /api/public/execution-readiness
  • /research/search-demand-action-map-2026
  • /api/public/search-demand-actions
  • /citation-packs
  • /answer-capsules
  • /freshness
  • /widgets

  Commands

  • npm run audit:seo-geo
  • npm run audit:technical-seo-controls
  • npm run audit:topical-authority-map
  • npm run audit:seo-geo-audit-evidence
  • npm run audit:execution-readiness
  • npm run audit:search-demand-actions
  • NEXT_PUBLIC_SITE_URL=https://visaatlas.org npm run build
  • npm run audit:rendered-seo

  Caveats

  • The release gate is local by default; run it with --include-liveness when network source checks are required.
  • The rendered audit checks representative artifacts; broad source audits cover the rest of the templates.
  • The audit is a regression gate, not a ranking guarantee.

• seo - instrumented

  Index quality and thin-page control

  index-quality

  The 500k/month target depends on a clean index; thin permutations can suppress genuinely useful pages.

  Target

  No known thin page classes indexed without unique sourced value.

  Cadence

  Every build and before adding new programmatic page cohorts.

  Evidence surfaces

  • /sitemap.xml
  • /compare
  • /professions
  • /from/indian/to/uk

  Commands

  • npm run audit:thin
  • npm run audit:noindex
  • npm run audit:longtail-provenance

  Caveats

  • A noindex guard is a trust-preserving choice when unique sourced value is not available.
  • Re-expansion should be earned by adding verified corridor, profession or comparison data.

• trust - instrumented

  Competitor wedge proof

  competitor-wedge-proof

  A Big4-style review should be able to inspect the competitor/wedge claim without relying on private strategy notes.

  Target

  The public positioning claim is backed by a reusable sampled benchmark with explicit caveats.

  Cadence

  Quarterly, or whenever major competitor surfaces materially change.

  Evidence surfaces

  • /research/immigration-source-transparency-benchmark-2026
  • /api/public/source-transparency-benchmark
  • /research/competitor-query-battlecards-2026
  • /api/public/competitor-query-battlecards
  • /answer-capsules
  • /api/public/answer-capsules
  • /mcp
  • /methodology

  Commands

  • npm run audit:source-transparency
  • npm run audit:competitor-query-battlecards
  • npm run audit:mcp
  • npm run audit:trust-surface

  Caveats

  • The benchmark is a public-surface sample, not a full forensic audit of competitors.
  • Not-observed means the feature was not visible in sampled public pages.

• seo - instrumented

  GSC cohort reporting

  gsc-cohort-reporting

  A site can grow impressions while losing CTR. Cohort reporting separates ranking, clickability and index-quality effects.

  Target

  Track clicks, impressions, CTR and average position by page cohort: calculators, data, research, core route reference pages, corridors, professions and comparisons.

  Cadence

  Monthly from a Search Console Pages export, and after major page-cohort launches.

  Evidence surfaces

  • /research
  • /research/ai-retrieval-readiness-scorecard-2026
  • /api/public/ai-retrieval-readiness
  • /research/ai-retrieval-task-index-2026
  • /api/public/ai-retrieval-task-index
  • /research/seo-geo-page-action-map-2026
  • /api/public/seo-geo-page-action-map
  • /research/quarterly-market-share-evidence-summary-2026
  • /api/public/quarterly-market-share-summary
  • /research/battlecard-search-experiments-2026
  • /api/public/battlecard-search-experiments
  • /research/competitor-query-battlecards-2026
  • /api/public/competitor-query-battlecards
  • /research/publisher-citation-kits-2026
  • /api/public/publisher-citation-kits
  • /research/authority-distribution-plan-2026
  • /api/public/authority-distribution-plan
  • /research/external-data-repository-mirrors-2026
  • /api/public/external-repository-mirrors
  • /research/knowledge-graph-entity-seeding-2026
  • /api/public/knowledge-graph-entities
  • /research/expert-reviewer-entity-program-2026
  • /api/public/expert-reviewer-entities
  • /research/reviewedby-schema-rollout-2026
  • /api/public/reviewedby-schema-rollout
  • /research/high-risk-page-review-acquisition-2026
  • /api/public/high-risk-page-review-acquisition
  • /research/p0-reviewed-page-refresh-2026
  • /api/public/p0-reviewed-page-refresh
  • /data
  • /calculators
  • /compare
  • /professions
  • /visibility-metrics

  Commands

  • npm run seo:gsc-cohorts -- gsc-pages.csv --json --period-start=YYYY-MM-DD --period-end=YYYY-MM-DD --property-label=gsc-domain-redacted --search-type=web
  • npm run audit:ai-retrieval-readiness
  • npm run audit:seo-geo-page-action-map
  • npm run audit:gsc-opportunities
  • npm run seo:gsc-opportunities -- gsc-queries.csv --json --period-start=YYYY-MM-DD --period-end=YYYY-MM-DD --property-label=gsc-domain-redacted --search-type=web
  • npm run seo:gsc-battlecard-experiments -- gsc-queries.csv --json --period-start=YYYY-MM-DD --period-end=YYYY-MM-DD --property-label=gsc-domain-redacted --search-type=web
  • npm run seo:gsc-experiment-outcomes -- gsc-experiment-outcomes.jsonl --json
  • npm run seo:surface-snapshots -- snapshots.jsonl --json
  • npm run seo:market-share-summary -- surface-summary.json --json
  • npm run seo:market-share-proof -- market-share-proof.json --json
  • npm run seo:repository-mirrors -- repository-mirror-log.jsonl --json
  • npm run seo:knowledge-graph -- kg-entity-log.jsonl --json
  • npm run seo:expert-reviewers -- reviewer-evidence-log.jsonl --json
  • npm run export:reviewedby-schema -- reviewer-evidence-log.jsonl
  • npm run seo:review-acquisition -- reviewer-acquisition-log.jsonl --json
  • npm run seo:p0-page-refresh -- p0-reviewed-page-refresh-log.jsonl --json

  Caveats

  • Runs locally on a user-supplied Search Console export with a declared period, redacted property label and search type; no private analytics are exposed in this public contract.
  • The public cohort taxonomy is now fully wired to the 48-prompt retrieval scorecard and 118-page SEO/GEO action map; live traffic proof still requires a reviewed private export.
  • Battlecard outcome claims require reviewed aggregate rows with matched baseline/comparison scope, redacted query rows and local-only raw export storage.
  • Stable-rank CTR should be analysed separately from rank movement.
  • Query-level opportunity backlogs are local working files and should not be committed.

• geo - instrumented

  AI referral attribution

  ai-referral-attribution

  AI visibility can improve before referral clicks arrive, and clicks can fall when answer engines cite the site directly. AI referral sessions are browser visits from AI products, not crawler hits.

  Target

  Separate AI crawler access, AI answer citations and AI referral sessions instead of treating them as one metric.

  Cadence

  Monthly.

  Evidence surfaces

  • /citation-benchmark
  • /research/canada-express-entry-category-tracker
  • /ai-agents
  • /api/public/visibility-metrics

  Commands

  • npm run seo:ai-referrals -- analytics-referrers.csv --json --period-start=YYYY-MM-DD --period-end=YYYY-MM-DD --property-label=ga4-redacted --bot-filtered
  • npm run kpi:citation-share
  • npm run seo:botlog -- logs.txt --json

  Caveats

  • Runs locally on a user-supplied analytics/referrer export with a declared period, redacted property label and bot-filter confirmation; private sessions are not published in this public dataset.
  • Crawler logs and citation-share logs are complementary, not interchangeable.
  • Answer engines may cite Visa Atlas without sending a referral click.

Cohorts