Know a pipeline broke before your CEO does.
For data teams of one to three people on Snowflake or Databricks. We scan your warehouse’s metadata for failed loads, stale tables and schema changes, then tell you in plain English what broke and what to fix first.
We do the work. You spend 15 minutes. No card needed. Or book a 15-minute call.
| Table | Last load | Rows vs 7-day avg |
|---|---|---|
| fct_orders | 31 h ago | no new rows |
| fct_payments | 9 min ago | −62% |
| dim_customers | 38 min ago | +1% |
| hubspot.deals | 12 min ago | type changed |
- B2fct_orders last loaded Tuesday at 2:10 a.m. The Stripe connector has failed every hour since because its API key expired, and the revenue dashboard still shows Tuesday.
- C3fct_payments loaded 1,140 rows against a 7-day average of 3,010. Yesterday’s dbt change added an inner join to a table that’s two days behind.
- C5deals.amount changed from NUMBER to VARCHAR. Two models cast it to a number and will fail on tonight’s run.
- B2New key in place. Loads are back to hourly.
- C3The join is a left join again. Row counts are back to normal.
- C5Both models handle the new type. Tonight’s run passes.
Sound familiar?
An executive found a stale dashboard before you did.
A load failed overnight, and nothing alerted you.
Someone changed a column upstream, and your models broke.
There are one to three of you, and no one has time to watch every table.
The longer a broken table goes unnoticed, the more decisions get made on wrong numbers.
In a 2023 survey of 200 data professionals, 74% said business users find data problems first all or most of the time (Monte Carlo, via Business Wire).
What you get, free
A full health check of your warehouse, done for you in 48 hours.
| What you get | Cost to you |
|---|---|
| 30 days of loads checked for failures, delays and stale tables | Included |
| Every row-count drop and schema change, with the likely cause | Included |
| A one-page report that ranks what to fix first | Included |
| Bonus: the SQL checks we ran, yours to rerun any time | Included |
| Bonus: a 15-minute call to walk you through the results | Included |
| Total | $0 |
No card. No contract. You keep the report and the SQL.
How it works
- Today, 15 minutes
- Run our SQL to create a read-only role. It sees load history, row counts and schemas, not your data.
- Within 48 hours
- We check 30 days of history and trace each problem to its cause.
- Day 2
- You get the report and the SQL. We walk you through it if you want.
After the scan, we watch it every day
- You get a Slack or email alert when a load fails, a table goes stale or a schema changes.
- Each alert says what broke, why, and what to do.
- A short weekly note in plain English for the rest of the company.
- Cancel any month.
Money-back guarantee
If the alerts aren’t useful in your first 30 days, we refund you in full.
Other ways to fix this
| Option | Cost | What you still have to do |
|---|---|---|
| Hire a contract data engineer | $30 to $100 an hour | Find one, onboard them and review their work |
| Buy a data monitoring platform | Custom pricing, through a sales call | Set it up and tune the alerts yourself |
| NumbersAgree | Free scan, then $300 a month | Read the alert and apply the fix |
Our promises
- We only see metadata like row counts and schemas, never the rows in your tables.
- We never change anything in your warehouse.
- A person checks every finding before you see it.
- You get a reply within one working day.
Questions
Is this software or a service?
A service. We run the scan and the daily checks for you. There’s nothing to install.
What access do you need?
A read-only role on metadata. On Snowflake, that’s the ACCOUNT_USAGE views for load, query and table history. On Databricks, it’s the system tables for jobs and table history. We send the exact SQL.
Can you see our data?
No. We look at row counts, timings and schemas, not the rows themselves.
We use dbt, Fivetran or Airflow. Does that matter?
No. They all leave a trail in the warehouse’s history. With read access to dbt Cloud or your orchestrator, we can also name the failing job, not just the table.
What if the scan finds nothing?
Then you have a report that shows your pipelines are healthy. That’s useful the next time someone asks.
Who is this not for?
Teams with a dedicated data platform or on-call rotation, and companies without a warehouse yet.
Get your free health scan
Tell us about your setup. We reply within one working day with the SQL for the read-only role.
Prefer to talk first? Book a 15-minute call.