GA4 Reporting: A Complete Guide

The most common GA4 pull isn’t sessions, it’s events. date, event_count, event_name is the single most-repeated field combination on the connector, ahead of any sessions-based pull. Alongside it, session_default_channel_group, conversions, and engaged_sessions round out what teams check first, GA4 users are reading behavior and outcomes, not just traffic volume.

Pull date, event_count, and event_name for my GA4 property over the
last 30 days. Which events are trending up or down week over week?

This works the same whether you land it in ChatGPT, Claude, or Copilot once GA4 is connected to Windsor.ai, or schedule the same pull into Looker Studio, Power BI, or a warehouse.

Events and key events

event_name and event_count are the base pull; conversions counts only the events marked as key events in GA4, and is_conversion_event flags which specific rows are key events within a broader event pull. Marking or unmarking an event as key only affects reporting from that point forward, it doesn’t rewrite history.

Pull conversions, event_count, and event_name for the last 30 days.
Which key events are growing fastest?

Acquisition

session_default_channel_group is one of the most-requested fields on the connector, an enumeration (Direct, Organic Search, Paid Social, Organic Social, Email, and more) built primarily from source and medium. session_source_medium gives the same information as one combined field and is a ChatGPT favorite specifically; the separate source and medium fields, by contrast, are requested by a much smaller share of ChatGPT users than of the connector’s traffic overall, most assistants asking about acquisition pull the combined field rather than the two halves.

Landing pages and content

landing_page is the page path from the first pageview in a session; page_path covers every page viewed, not just the entry point. screen_page_views counts both, repeated views included. Pulling page path with screen page views and sessions together is one of the most-repeated combinations on the connector, the standard content-performance pull.

Engagement

engaged_sessions counts sessions that lasted over 10 seconds, included a key event, or had two or more screen views. engagement_rate is engaged sessions divided by sessions; bounce_rate is its inverse. Engagement rate is requested by roughly half again as many teams as bounce rate, GA4’s own preferred framing has mostly displaced the old bounce-rate habit.

Users

active_users is GA4’s primary user metric, and users is a direct alias for it. totalusers counts everyone who logged at least one event, a broader definition than active users. ChatGPT users pull active_users far more often than totalusers, relative to their overall traffic share, the opposite skew from most other fields on this connector. See The Day ChatGPT Changed How It Counts Your Users for the full story behind that flip.

Ecommerce

purchase_revenue, transactions, ecommerce_purchases, and item-level fields like item_name and item_revenue are heavily pulled in their own right, and they’re also where GA4’s field-compatibility errors concentrate most. date, purchase_revenue, sessions, transactions and item_name, item_revenue, items_purchased are the two combinations that reliably work together.

Paid joins: GA4 and Google Ads

A large majority of GA4 accounts also pull Google Ads in the same week, more than any other cross-connector pairing on this platform. session_google_ads_campaign_name and session_manual_campaign_name join the two on campaign name for a blended view. For a full CPA model joining the two in a warehouse, see Google Ads CPA Tracking in BigQuery (with GA4), and for the reporting side, Google Ads Reporting and Analytics: A Complete Guide.

Search Console join

More than a third of GA4 accounts also pull Search Console in the same week. The organic_google_search_clicks, organic_google_search_impressions, and organic_google_search_average_position fields bring Search Console’s own metrics into a GA4 pull directly, without a separate report. See Connect Google Search Console to Claude to set up the Search Console side.

Destinations

Power BI is the single largest named destination for GA4 data by volume, more than any other connector’s Power BI traffic. AI assistants, Claude ahead of ChatGPT, and Google Sheets follow. A meaningful share of GA4 traffic also comes through direct API clients and automation platforms like n8n and custom scripts, teams building their own pipeline rather than using a named destination app.

Field names that don’t exist

A recurring source of errors: guessed field names that don’t match Windsor’s actual ids. total_users is totalusers, new_users is newusers, device_category is devicecategory, session_source and session_medium are session_source_medium, or source and medium separately, and total_revenue is totalrevenue. This single pattern accounts for more failed GA4 pulls than any other error family on the connector. The full lookup table is in the GA4 “Unexpected field(s)” error guide.

Limits worth knowing

GA4 allows at most 10 metrics in a single request; requesting an 11th fails the whole pull rather than truncating it. Full details, including date-preset and filter rules, are in the GA4 request limits guide. A separate, larger error family covers fields that simply can’t be requested together, transactions, event counts, and ecommerce fields are the most common triggers, each needing its own request rather than one combined pull; see the “fields cannot be queried together” guide for the fix per case.