BigQuery Documentation

Reach Dog commerce-map data for analytical SQL workflows

Reach Dog maintains a commerce map that connects product listings, semantic product neighborhoods, buyer-language questions, analyzed keywords, attributes, buyer dimensions, and market signals. The BigQuery datasets expose parts of that structure for analysts and data buyers.

The platform authority figures are separate from warehouse table row counts: Reach Dog has analyzed 101.3 million keywords, maintains 70.5 million modeled natural-language buyer questions, and maps more than 360 million product listings into 29.6 million semantic product neighborhoods. The question layer models how buyers describe needs; it is not presented as harvested query logs.

Scope note: Registry totals describe the wider Reach Dog platform. Do not treat them as the row count of a single BigQuery table or add them together as if they were equivalent units.

Project and datasets

The current warehouse project is concise-ivy-489905-i8.

DatasetPurpose
TAXONOMYCore commerce-map tables and product-to-buyer-language relationships
enriched_dataAnalytical lens views built over the core tables

Physical table counts are intentionally omitted here. Local schema manifests and dated warehouse snapshots do not establish one stable current public total.

Core table reference

These are the principal tables used by current analytical code. Their names are literal schema identifiers.

TablePurpose
TITLE_CANONICALSSemantic product-neighborhood identifiers, representative titles, and available enrichment tags
PRODUCT_BRIDGEProduct-to-neighborhood relationships
INTENT_PRODUCTSAnalyzed keyword records and available search-volume fields
INTENT_PRODUCT_MATCHESKeyword-to-product-neighborhood relationships with semantic relevance
KEYWORD_FINANCIALSAvailable keyword market signals such as search volume, CPC, and competition
QUESTIONSModeled buyer-question phrasings mapped to product neighborhoods
FINANCIAL_METRICSAvailable pricing, demand, competition, and market-position signals at the neighborhood level
PRICE_HISTORY_EVENTSDated observed price-change events
WEEKLY_DIMENSIONSDated dimension aggregates; this table name does not imply that every platform layer updates weekly

Relationship model

Core join path
INTENT_PRODUCTS -> INTENT_PRODUCT_MATCHES -> TITLE_CANONICALS -> PRODUCT_BRIDGE KEYWORD_FINANCIALS adds available keyword market signals. QUESTIONS adds modeled buyer-language phrasing. FINANCIAL_METRICS adds available neighborhood-level market context.

This is a technical graph model inside the warehouse. Reach Dog's public product description remains the commerce map: product reality connected to buyer language across the open market.

17 analytical lens views

The canonical BigQuery view definition contains 17 views in enriched_data. They provide reusable analytical starting points over the core tables.

voice_search_lensgaps_lenspricing_lensghosts_lensproduct_scorecard_lensmarket_overview_lensdemand_map_lensmarket_audience_lensaudience_content_lenscompetitive_landscape_lenscompetitor_pricing_lensmodality_strategy_lensopportunities_lenscannibalization_lenscontent_calendar_lenscatalog_brief_lensppc_campaign_lens

Five buyer dimensions

The current buyer-dimension model is What, Where, When, Who, and Why or benefit. Category and other warehouse tags may coexist with these dimensions, but category is not one of the five buyer dimensions.

DimensionMeaning
WhatWhat the product does or the use case it serves
WhereThe situation, setting, or context in which it is used
WhenThe timing, season, or recurring moment
WhoThe buyer, recipient, or demographic context
Why / benefitThe benefit or outcome the buyer is trying to get

Measured and inherited demand signals

Where present, SV represents measured search volume. PARENT_SV is inherited from a parent keyword. They are not the same measurement and should not be described or aggregated as though both were directly measured for the child term.

Coverage varies by field and data layer. This page does not publish a blanket percentage for the share of keywords carrying demand signals.

Freshness varies by data layer

The warehouse receives scheduled refresh activity, while freshness varies by data layer. Dimension tables, keyword signals, product relationships, question mappings, price history, and historical aggregates do not share one universal update time. Analysts should use the applicable dated fields and delivery metadata for the layer they query.