Free Analysis for BigQuery
Find out how much of your BigQuery spend might be wasted - before you set anything up. The free analysis is built on your own workloads, not a generic benchmark. It looks at how each of your jobs is billed today and finds the cheapest way to run it:
- On-demand or reservations - which workloads cost less billed by data scanned (on-demand), and which cost less on slots (reservations)
- Standard edition reservations - which workloads could run on a Standard edition reservation instead of Enterprise
It then quantifies your potential savings with Yuki in dollars.
How It Works
- Export 30 days of metadata - run a few read-only queries yourself. No query text, no table contents.
- We replay your real workload - Yuki's optimizer runs against your actual usage to show how your environment would behave with Yuki in place.
- Get your report by email - see where your compute goes and the dollar value of every fix.
It takes about 3 minutes of your time. No credit card, no commitment. All data is automatically deleted after 30 days, and Yuki is SOC 2 and GDPR compliant.
Step 1 - Start Your Free Analysis
You'll need: the free analysis link and secret key from the Yuki team, your BigQuery region, what you pay for BigQuery compute, and permission to read job history.
Which permissions do I need?
| To export | You need |
|---|---|
| All projects at once (Automated Export) | roles/bigquery.resourceViewer at the organization or folder level |
| Specific projects | The bigquery.jobs.listAll permission on each project |
| Reservations usage (if you use reservations) | bigquery.reservations.list (BigQuery Resource Viewer) on your reservations administration project |
- Open the free analysis link
- Enter your Secret Key and business Email
- Under Which platform should we analyze?, select BigQuery
- Select your BigQuery region and click Start
- Choose how to share your data - Automated Export (recommended) or Manual Upload (see Step 3)
Step 2 - Enter Your BigQuery Costs
Yuki asks How are you billed for BigQuery compute? Select all that apply:
| Billing model | Field | Default |
|---|---|---|
| On-demand - billed by data scanned | Current cost per TiB (USD) | $6.25 |
| Reservations (capacity pricing) - billed by slot-hours through a BigQuery edition | Current cost per slot-hour (USD) | $0.04 |
Enter your effective price, including any discounts or commitments, and click Save next to each field. Yuki needs at least one billing model to continue.
Step 3 - Share Your Data
Both options run the same queries and share the same metadata. Yuki shows up to 3 queries, depending on your billing model:
| Query | What it exports |
|---|---|
| Jobs Metadata | Job statistics for the last 30 days |
| Jobs Timeline | Slot usage over time for the last 30 days |
| Reservations Timeline | Reservation usage and slot allocation. Shown only if you selected Reservations. |
- Automated Export (recommended)
- Manual Upload
The query results are exported straight to a private Yuki storage bucket - no file handling.
- Choose which projects to export:
- All projects - one run per region exports every project with BigQuery usage. Requires organization- or folder-level access.
- Specific projects - enter a Project ID, then run the queries for that project. Run Jobs Timeline in each of your ETL projects.
- Copy each query shown on the page, run it in your BigQuery console, and click I Ran This Query
- Click Continue
If the All projects queries fail with Access Denied, you don't have organization-level access. Switch to Specific projects, which only needs access to each project.
Run the queries yourself, export the results, and upload the files to Yuki. For each query on the page:
- Copy the query and run it in your BigQuery console. For Jobs Metadata, enter the Project ID first and run the query in each of your BigQuery projects.
- Export the results and download them as Parquet files
- Attach the files under the query and click Save. You can attach one or more
.parquetfiles, or a single.zipor.gzfile - but not both.
When every file is saved, click Continue.
What data does Yuki receive?
Both options share the same metadata from BigQuery's INFORMATION_SCHEMA:
- Jobs metadata - job type, timings, bytes processed and billed, slot usage, status, and labels. Users are identified only by the part of their email before the
@. - Jobs timeline - slot usage per job over time
- Reservations timeline - reservation usage and slot allocation (only if you use reservations)
Query text is never collected. Instead, Yuki receives a one-way hash of each query's text, so it can recognize repeated queries without seeing them. All data is automatically deleted after 30 days.
Step 4 - Get Your Report
You'll see Thank You - You're All Set! Yuki's optimization experts review your results to put together the most accurate report for your setup, and email it to you when it's ready.
Questions? Contact support@yukidata.com.