How to Set Up GA4 BigQuery Export (Step-by-Step).

One of the most powerful features of GA4 is the free BigQuery export — yet a significant share of properties we audit don't have it turned on. If you're still relying solely on the GA4 UI for analysis, you're leaving your most valuable data on the table. BigQuery export gives you access to the raw, unsampled, event-level data that powers everything from advanced attribution modeling to machine learning predictions.

This step-by-step guide covers everything you need to know about setting up, configuring, and using GA4 BigQuery export in 2025, including realistic cost estimates, essential starter queries, and common pitfalls to avoid.

Skip the manual check: the BigQuery export audit verifies streaming-vs-daily mode, sampling-threshold risk against your 1M-events/day cap, and whether your 14-month retention cliff is already escaping into BigQuery — in 60 seconds, no Cloud Console required.

Why BigQuery Export Matters

The Limits of the GA4 Interface

The GA4 interface is designed for quick, high-level analysis. But it comes with significant limitations that affect data accuracy and analytical depth:

• Sampling: GA4 applies data sampling to explorations and reports that exceed certain thresholds. BigQuery data is never sampled — you always work with 100% of your events.
• Thresholding: As we covered in our article on GA4 thresholding, Google Signals triggers data hiding in the UI. BigQuery receives the complete, unthresholded dataset.
• Data limits: GA4's standard reports have limitations on the number of dimensions, metrics, and date ranges you can combine. BigQuery has no such restrictions — you can write any SQL query against the full dataset.
• Data retention: GA4 UI retains detailed event data for at most 14 months. BigQuery stores data indefinitely (as long as you're willing to pay the storage costs, which are minimal).
• Join capability: BigQuery allows you to join GA4 data with CRM data, advertising data, product catalogs, weather data, or any other dataset. This enables advanced analyses that are impossible in the GA4 interface.

Setup and Configuration

Prerequisites

Before setting up the export, ensure you have the following:

1. A Google Cloud Platform (GCP) project: If you don't have one, create a free project at console.cloud.google.com. You'll need to add a billing account, but the BigQuery free tier covers most small-to-medium sites.
2. Editor or Admin access to both the GA4 property and the GCP project.
3. BigQuery API enabled in your GCP project (it's usually enabled by default).

Setup in 5 Minutes

1. Open your GA4 property and go to Admin → BigQuery Links.
2. Click Link and select the GCP project where you want the data exported.
3. Choose your export frequency: Daily (recommended for most sites) or Streaming (for real-time needs).
4. Select the events you want to export. We recommend all events initially — you can always filter in your queries later, but you can't retroactively export events you didn't include.
5. Choose a data location (e.g., US or EU) — this should match your GCP project's default region for cost optimization.
6. Click Submit. Wait 24-48 hours for the first tables to appear in your BigQuery project.

Working with the Data

Understanding the BigQuery Schema

GA4 exports data into tables with a specific naming convention and structure:

• Daily tables: Named events_YYYYMMDD (e.g., events_20250228). One table per day, created after the day is complete.
• Intraday tables: Named events_intraday_YYYYMMDD. These contain streaming data for the current day and are replaced by the daily table once the day is finalized.

Each row in the table represents a single event. The schema uses nested and repeated fields (RECORD and ARRAY types), which is different from flat tabular data. Key fields include:

• event_name — The name of the event (e.g., page_view, purchase).
• event_params — A repeated RECORD field containing all event parameters as key-value pairs.
• user_properties — A repeated RECORD field containing user-scoped properties.
• user_pseudo_id — The client ID (equivalent to the GA cookie).
• geo, device, traffic_source — Nested records with geographic, device, and acquisition data.

Essential Starter Queries

Here are three queries every analyst should run on their first day with BigQuery export (see our complete BigQuery export guide for the full schema reference and more queries):

1. Daily active users by event:

SELECT event_name, COUNT(DISTINCT user_pseudo_id) AS users
FROM `project.dataset.events_*`
WHERE _TABLE_SUFFIX BETWEEN '20250201' AND '20250228'
GROUP BY event_name ORDER BY users DESC;

2. Unnesting event parameters:

SELECT event_name,
  (SELECT value.string_value FROM UNNEST(event_params) WHERE key = 'page_location') AS page,
  COUNT(*) AS events
FROM `project.dataset.events_*`
WHERE _TABLE_SUFFIX = '20250228'
GROUP BY 1, 2 ORDER BY events DESC LIMIT 20;

3. Revenue by traffic source:

SELECT traffic_source.source, traffic_source.medium,
  SUM(ecommerce.purchase_revenue) AS revenue
FROM `project.dataset.events_*`
WHERE event_name = 'purchase'
  AND _TABLE_SUFFIX BETWEEN '20250201' AND '20250228'
GROUP BY 1, 2 ORDER BY revenue DESC;

Cost and Common Mistakes

Cost Reality Check

BigQuery pricing has two components: storage and query processing. Here's what to expect:

• Storage: $0.02/GB/month for active storage. A site with 1M events/month generates roughly 1-3 GB of data — costing $0.02-0.06/month. Even at 10M events/month, storage costs are typically under $1/month.
• Queries (on-demand): The first 1 TB of queries per month is free. After that, it's $6.25/TB. Most analytics queries scan well under 1 TB per month.
• Streaming export: Adds real-time data availability but costs approximately $0.05 per 200MB of streamed data. Only enable if you genuinely need sub-hour data latency.

For most sites under 10M events/month, the total BigQuery cost is under $5/month — often effectively free within the free tier. This makes BigQuery one of the highest-ROI analytics investments you can make.

Common Mistakes to Avoid

• Not enabling export early enough: BigQuery export is not retroactive. It only captures data from the day you enable it forward. Every day without export is data you'll never get back.
• Choosing the wrong data location: Select a BigQuery region that matches your GCP project and is close to your team. Cross-region queries incur additional costs.
• Not understanding UNNEST: GA4's nested schema requires the UNNEST function to access event parameters. This is the number one source of confusion for analysts new to BigQuery.
• Running expensive queries: Always use _TABLE_SUFFIX filters to limit the date range of your queries. Without date filtering, queries scan all historical data, which can become costly as your dataset grows.

Key Takeaways

• BigQuery export provides raw, unsampled, unthresholded GA4 data — essential for serious analytics work.
• Setup takes 5 minutes and costs are negligible for most sites (under $5/month).
• Enable export as early as possible — it's not retroactive.
• Learn the UNNEST pattern for querying event parameters — it's the key to unlocking BigQuery's full potential.
• Use daily export for most cases; streaming export only when sub-hour latency is required.

Frequently Asked Questions

Why export GA4 data to BigQuery?

The GA4 interface applies sampling to large datasets, hides data through thresholding when Google Signals is active, and limits how many dimensions and date ranges you can combine in a single report. BigQuery export bypasses all of these constraints. You get 100% of your raw, event-level data in a SQL-queryable format that you can join with CRM records, advertising data, or any other source. It is also the only mechanism for retaining event-level data beyond GA4's 14-month maximum retention window.

Is GA4 BigQuery export free?

The export link itself is free — Google does not charge to send your GA4 data to BigQuery. However, BigQuery storage and query costs are billed through your Google Cloud project. Storage runs at $0.02 per GB per month, and the first 1 TB of query processing per month is free under the on-demand pricing model. For sites with fewer than 10 million events per month, total costs typically stay well under $5 per month. Streaming export adds a per-volume charge and is the only component that meaningfully affects cost at smaller scale.

What is the difference between daily and streaming BigQuery export?

Daily export creates one finalized table per day (named events_YYYYMMDD) after that day's data has been fully processed by GA4. Data typically arrives within a few hours after midnight. Streaming export writes events to an intraday table (events_intraday_YYYYMMDD) within minutes of collection. The intraday table is replaced by the finalized daily table once the day closes. Daily export is sufficient for the vast majority of use cases. Streaming export is appropriate only when your analysis requires near-real-time data, such as same-day operational dashboards or fraud detection workflows — it adds incremental cost and schema complexity without benefit for standard retrospective analysis.

How long does GA4 BigQuery export history go back?

BigQuery export is not retroactive. It captures events from the day the link is created and forward. Google does not populate historical tables for periods before the export was enabled. If you enable the export today, tomorrow's query on last year's data will return no results. For this reason, enabling the export as early as possible — ideally on the day you implement GA4 — is the single most important data infrastructure decision you can make. There is no mechanism to recover events that were never exported.

How do I query GA4 data in BigQuery?

GA4 data in BigQuery uses a nested schema. Each row represents one event, and event parameters are stored in a repeated RECORD field called event_params. To read a specific parameter value, you use BigQuery's UNNEST function: SELECT (SELECT value.string_value FROM UNNEST(event_params) WHERE key = 'page_location') AS page FROM `project.dataset.events_20250228`. Always filter by _TABLE_SUFFIX when querying across multiple dates using the wildcard table syntax (events_*), otherwise your query will scan the full history and incur unnecessary cost.

What format does GA4 data use in BigQuery?

GA4 exports data in a nested, columnar format using BigQuery RECORD and ARRAY types. Each row is one event. Top-level fields like event_name, user_pseudo_id, and event_timestamp are flat scalars. Fields like event_params, user_properties, items, geo, device, and traffic_source are nested records or repeated records. This structure is more expressive than flat tabular data but requires familiarity with UNNEST and dot-notation field access to query effectively. Google publishes a full schema reference in the GA4 documentation.

Can I backfill historical GA4 data in BigQuery?

No. Google does not provide a mechanism to export historical GA4 data to BigQuery retroactively. If you enabled GA4 tracking before enabling the BigQuery export, those earlier events exist only in GA4's aggregated reports and cannot be recovered at the event level. Some third-party tools claim partial backfills by re-processing GA4 API export data, but these reconstructions are not equivalent to the raw event-level export and have significant gaps. The only reliable approach is to enable the export before you need the historical data.

Does the GA4 BigQuery export count toward GA4 limits?

The BigQuery export itself does not consume GA4 hit quotas or affect data collection limits. However, properties on the free tier that exceed 1 million events per day will have the excess events exported to BigQuery but not processed in the standard GA4 reports — the export is your safety net in this scenario. Streaming export has its own quota limits within Google Cloud (governed by BigQuery's streaming insert quotas), but these are rarely relevant for standard analytics workloads. The export runs server-side and does not add any load to your website or tag implementation.