BigQuery Pricing: What It Costs, the Free Tier and Worked Examples

BigQuery charges for two things: storing data and processing it. There are no per-user licences and no servers to rent. For queries there are two models:

• On-demand: you pay for the bytes each query reads. This is the default.
• Capacity (editions): you pay for slots, which are virtual CPUs, by the hour as you use them or through one- or three-year commitments.

The figures on this page come from the Google Cloud pricing page as checked in October 2026, in US dollars, for the US and EU multi-regions. If you pay in another currency, Google applies the prices listed in your currency on its Cloud SKUs page.

The price depends on where your data lives. The US and EU multi-regions have the prices above. Some single regions cost more: in Madrid (europe-southwest1), on-demand queries are $7.8125 per TiB and active storage $0.02875 per GiB a month; in Belgium (europe-west1), $7.50 per TiB, with storage at the same price as the EU multi-region.

If your data has to stay in the European Union, the EU multi-region already does that. A single region makes sense when there is a specific reason for the data to sit in one country.

It matters most when you set up the GA4 export: changing the location later means deleting the link and recreating the dataset. How the GA4 BigQuery export works.

A few rules from the pricing page that explain a lot of bills:

• Queries that return an error and queries served from the cache are not charged.
• Each query is billed for at least 10 MB, plus at least 10 MB for each table it references.
• You pay for the columns a query reads. Adding LIMIT does not reduce what is read: a SELECT * with LIMIT 10 is billed in full.
• Batch loading is free when it uses the shared slot pool. What you pay for afterwards is storing the data.
• Cancelling a running query can cost as much as letting it finish.

Partly. The BigQuery free tier gives you 1 TiB of queries and 10 GiB of storage each month, and at small volumes it is easy to stay within it.

If you only want to try it, the BigQuery sandbox works without a credit card or a billing account. As of October 2026, Google's documentation sets these limits:

• 10 GiB of storage for the lifetime of the sandbox, which is not given back when you delete data.
• 1 TiB of queries a month, the same free tier as everyone else.
• Tables, views and partitions expire automatically after 60 days.
• No streaming, no DML statements (INSERT, UPDATE, DELETE, MERGE) and no Data Transfer Service.

It is fine for learning SQL or seeing how GA4 data arrives. It is not a way to keep a history: after 60 days the tables are gone.

Three cases to give you a feel for it. They are calculations with the US and EU multi-region prices, rounded, not anyone's actual bill.

A company website with GA4

Google estimates that 1 GB is about 600,000 GA4 events. A site with 300,000 events a month adds half a GB a month; after two years it is still under 12 GB, nearly all of it inside the free 10 GiB. If reports read only the columns they need, queries stay far below the free TiB. Typical cost: zero or close to it. For what changes on a busy property, see the GA4 BigQuery export limit and its costs.

A dashboard on raw data

A dashboard whose queries read 50 GiB every time it is opened, with no cache, opened 40 times a day over 22 working days: 50 × 40 × 22 = 44,000 GiB, about 43 TiB a month. Take away the free TiB and about 42 TiB remain at $6.25: around $260 a month.

The same dashboard on a 1 GiB summary table reads 880 GiB a month and stays inside the free tier. The difference is not BigQuery, it is how the model is built.

A mid-sized data warehouse

500 GiB stored, half of it untouched for more than 90 days: 250 × $0.02 + 250 × $0.01 = $7.50 a month, minus what the free 10 GiB cover. Storage is rarely what pushes the bill up; queries are.

They are often compared because both turn up in every business intelligence project, but they do different jobs:

• BigQuery stores and computes. It is the data warehouse: where GA4, Google Ads, the CRM or the ERP land, where they are joined, and where tables with the company's definitions are prepared. What a data warehouse is.
• Power BI presents. It is the reporting tool: it connects to a source, defines the measures and draws the dashboards the team uses.

The usual setup uses both. Power BI has a Google BigQuery connector and works in two ways: import, which loads the data on each scheduled refresh, or DirectQuery, which queries BigQuery whenever someone opens or filters a report. With DirectQuery every click can cost money, so it should read summary tables.

The same goes for Looker Studio or Tableau. If you are choosing the reporting tool, we compare the three here: Power BI vs Looker Studio vs Tableau.

At low volumes, on-demand is usually the cheapest option: the first TiB each month is free and there is nothing to reserve. Editions start to make sense when query spend is high and steady, or when you need guaranteed capacity.

It is a decision to make with data: BigQuery records the bytes billed and the slot time of every query in INFORMATION_SCHEMA.JOBS, and a couple of months of history are enough to compare one model with the other. It is part of what we review in our BigQuery consulting.