Organic acquisition context
Use optional target-bound Google Search Console aggregates to understand which organic search landing pages merit investigation. This supplements the on-site funnel; it does not supply session conversion rates, prove a cause, or establish experiment assignment.
Bind and read safely
Start from the configured storefront and any supplied property identity. Verify that the exact URL-prefix or Domain property covers the target origin and intended market before reading search data. For a Domain property spanning multiple hosts, constrain page evidence to the configured storefront and record the scope. Verify access using source metadata or the exact site's read operation; do not choose the store from an ambient connection or query another property's business data to find a convenient target.
Use the host's tool discovery for permitted read-only metadata and Search Analytics operations, or an authorized aggregate report artifact with equivalent provenance. With Composio, the toolkit is google_search_console; other hosts may expose equivalent operations under different names. Discover schemas instead of assuming an adapter or tool slug. If the host blocks a required read, record that capability gap without proxying around its guard.
For a host enforcing a read-only review boundary, verify from host-provided capability/permission evidence or non-mutating tool/schema inspection that filtering recognizes the actual operations. A toolkit name containing SEARCH is not proof that its add, delete, or submit actions are excluded. Do not use the direct connector in that runner when operation-level filtering cannot be verified; consume an authorized aggregate artifact instead and record the connector limitation. Never invoke a mutation to test whether it is blocked. An ordinary execution-capable host still follows the skill's read-only operation boundary and its own permissions.
Never add/delete a site, submit a sitemap, change property permissions or configuration, request indexing, or repair tracking through this report. User-authorized account connection is a separate setup handoff, never an operation inside guarded report evaluation. Keep credentials and personal identifiers out of queries and artifacts. Query the smallest responsible aggregate scope and avoid exposing sensitive or overly narrow query rows.
Preserve source definitions
Record property/type, target/access, extraction time, exact comparison dates, search type, dimensions/filters, aggregation and coverage. Default to finalized web-search data. Search Analytics dates use Pacific time; distinguish store/GA4 timezones rather than implying matching windows.
Preserve source clicks, impressions, CTR, position and units. CTR is clicks/impressions, returned as a fraction; label conversion to percent. Use actual source aggregates/supported weighting, not invented averages of row CTR/position.
Inspect reporting delay and incomplete-data metadata when available. Missing, anonymized, filtered, or omitted rows are not zero. Pagination and row limits do not guarantee a complete query population: Search Analytics can return only top rows. Describe the actual scope and do not label a bounded page/query extract exhaustive.
Relate acquisition to the funnel
Normalize landing URLs using verified hostnames, redirects, and canonical/market context. Strip tracking parameters and redact private query values while preserving functional distinctions. Compare search trends within like-for-like source definitions, dates, search type, device, country, and query intent. Record position, branded intent, seasonality, content changes, traffic mix, and reporting limits as alternative explanations.
Use material search reach to select landing pages for direct browser investigation. Relate those pages to separately measured Shopify or GA4 organic segments only when their route, market, and window definitions support the comparison. Do not treat search clicks as sessions, divide a source's purchases by another source's clicks, join individual people across sources, or infer a conversion defect from CTR or position alone.
Store acquisition metrics, query definitions, observations, and uncertainty in supporting evidence for the relevant existing funnel Check, usually funnel.landing-to-collection. This evidence can remain visible even when that rate is blocked. Keep the five Data keys and window, segment, and source dimensions unchanged; do not emit search metrics under an on-site rate key or introduce a sixth Check, Data key, Summary, or Finding. Use the host-supplied schema's permitted evidence fields for JSON and the corresponding Check evidence for Markdown.
Missing Search Console access never blocks an otherwise supported funnel rate. If Search Console is the only available source, preserve useful acquisition evidence but mark all five on-site Checks individually blocked for missing Shopify/GA4 session/outcome evidence. Do not report a measured funnel or fabricate rates. An Issue still needs the existing funnel contract's measured leak, direct storefront evidence, one concrete change or experiment, metric, guardrails, and retest.
Sources
Verified 2026-10-22; connector availability and tool permissions must still be discovered on the executing host.