Platform
Connect BigQuery, read-only
Pick project or organization, create the read-only role and key, test it and run the first scan.
Google CloudAvailable
BigQuery cost optimization, from metadata alone
Connect a read-only service account, scan your query history, and get a dollar figure per finding in under 10 minutes. 23 heuristic rules, each documented with the threshold that fired.
Execution-plan truth
Partition and CTE rules read job_stages READ operators, not a grep of the SQL text — so a WHERE clause that never reached storage still counts.
23 rules, all documented
No opaque score. Each finding names the table or query, the threshold, a sample job, and the estimated monthly savings.
Reservations without guesswork
JOBS_TIMELINE and RESERVATIONS_TIMELINE power on-demand vs Editions, idle baselines, contention, and PENDING share — fail-soft if the admin project is out of reach.
Attribution that names a person
dbt, Looker, Connected Sheets, Airflow, Dataform, scheduled queries, and console jobs are classified from job ids, job labels and comment headers.
What the scanner checks today
The BigQuery-native cost scanner
Missing partition filter
READ stages with no storage-level predicate on a PARTITION BY table.
SELECT * over-fetch
Wide tables scanned with star, billed far beyond the columns a dashboard used.
Compute pricing model
On-demand bytes vs Editions slot-hours at list, on the same 30-day window.
Storage billing type
Logical vs physical dataset billing, plus stale unused heaps.
Reservation health
Idle baseline, slot contention, and PENDING share from per-minute timelines.
MERGE / CTE / clustering
Wholesale target rewrites, CTE re-reads, and CLUSTER BY gaps.
Metadata only. Table contents stay put.
What we ask for
bigquery.jobs.create
Run INFORMATION_SCHEMA queries
bigquery.jobs.listAll
Read every user's job history
bigquery.datasets.get
Read dataset metadata
bigquery.tables.get
Read table metadata
bigquery.tables.list
List tables in a dataset
bigquery.reservations.list
List slot reservations
resourcemanager.projects.get
Read this project's metadata
bigquery.readsessions.create
Open a Storage Read session
bigquery.readsessions.getData
Stream rows from a Storage Read session
What we never ask for
roles/bigquery.dataViewer
The role that would grant access to table contents. Never requested.
See it on BigQuery
The product tour and short walkthroughs, recorded on a real account. Each one has a voice-over.
Product tour
SlashPeak in under 90 seconds
The whole loop: see where the spend goes, rank the savings, and let the Cognis AI DBA ship the fix. With voice-over.
Inside · 1:25
- Cost Analytics
- Savings Findings
- Cognis AI
SlashPeak Analytics
Reading your spend dashboard
Month-to-date spend, forecast, pricing model and spend composition.
SlashPeak Findings
Working through savings findings
How findings are ranked, what the evidence shows and how they close.
Cognis AI DBA
Ask Cognis about your bill
Plain-English questions answered from your own cost data.
Cognis AI DBA
Chat with Cognis AI
Start from a suggested question, ask follow-ups, jump to the right page and keep every thread.
SlashPeak Analytics
Reservations and slot usage
Spot idle baselines and contention before you resize a reservation.
BigQuery cost notes
All articles for this warehouse →
Why dbt MERGE jobs scan your whole BigQuery table
A MERGE touching only yesterday can still rewrite the whole target. SlashPeak catches the missing partition filter from the plan, not the YAML.
17 min read
How BigQuery clustering complements partitioning
A DATE partition still reads every block in that day without CLUSTER BY on account_id. Clustering makes the partitions you paid for skip blocks too.
14 min read
Ad-hoc queries and BigQuery reservation contention
INTERACTIVE is a latency SLA for one workload, not a default. Ad-hoc humans hogging a shared reservation queue the finance dashboard behind console SQL.
13 min read
BigQuery CTEs that scan the same table twice
BigQuery does not always materialize a CTE. Two references can mean two reads of the same 4 TB table. The SQL looks DRY; the execution plan is the invoice.
14 min read
BigQuery slots explained: usage, contention, and cost
A slot is compute, a slot-hour is a billing line, and contention is demand sitting on the ceiling. Here is how BigQuery capacity works.
20 min read
Take control of your warehouse bill
Understand, reduce and govern your cloud data spend with SlashPeak — starting today.
- 30-day free trial
- No credit card
- Read-only access
- Set up in minutes