// Data Platform Health Check
Know exactly what your data platform is costing you. In 10 days.
A fixed-scope, fixed-fee audit of your Databricks or cloud data platform. We find the misconfigurations, dead pipelines, and cost leaks, then back the findings with a guarantee.
8+ years designing production data platforms across banking, manufacturing, retail, and RegTech.
› provision access --read-only
✓ day 1–2 · telemetry collection live
› audit --clusters --jobs --storage --code
✓ findings quantified from actual telemetry
› deliver report --prioritized --day-10
● savings < fee → you pay nothing
The Offer
One priced engagement. No hourly billing, no scope surprises.
Single workspace, single cloud account
Up to 100 scheduled jobs/pipelines, 1 Databricks workspace, 1 cloud account
Multi-workspace or multi-cloud
Additional workspace: +$6,000 each
The fee is credited in full toward any remediation engagement started within 90 days of report delivery.
// The Guarantee
If the annualized savings identified in the report (calculated from actual cluster telemetry and job run history) don't exceed the cost of the audit, you pay nothing.
Scope
In Scope
- ✓1 Databricks workspace
- ✓1 cloud account (Azure or AWS)
- ✓Up to 100 scheduled jobs/pipelines
- ✓Associated cloud storage and orchestration
- ✓Git repo review for pipeline code
Add-ons (Priced Separately)
- +BI/reporting layer review
- +Data warehouse review (Snowflake, Synapse, BigQuery)
- +Application-layer review
- +Compliance-certification prep (SOC 2, ISO 27001)
The 10-Day Timeline, Honestly
10 business days is achievable, but it assumes three things on your side:
- 1. Read-only access is provisioned within 48 hours of kickoff.
- 2. One technical point of contact is available and responds same-business-day.
- 3. Your team's total time commitment is roughly 2-3 hours across kickoff, a mid-check-in, and the executive presentation.
If access is delayed, the timeline shifts day-for-day. We start the clock when access starts, not when the contract is signed.
Day-by-Day Methodology
Sample Findings
Anonymized, representative of what a typical assessment surfaces.
Misconfigured autoscaling
$4,200/moA cluster policy scaling on the wrong metric, running idle capacity around the clock.
Unoptimized shuffle
45 min → secondsA 3TB daily ETL job spilling to disk on every run from a skewed join key.
17 zombie scheduled jobs
$2,100/moJobs still running against tables deprecated over a year ago.
Un-Z-ordered fact table
12 min → 8 secA 40TB fact table with no clustering, forcing full scans on point queries.
// Cost model
Where the Savings Come From
The sample findings above are instances. This is the systematic sweep behind them: six levers, checked on every audit, quantified from your telemetry.
Cluster right-sizing & autoscaling
Policies sized to the workload's actual profile, not to worst-case guesses that idle around the clock.
Runtime & Photon
Outdated runtimes and unused Photon leave paid-for performance on the table. Upgrades are quantified before they are recommended.
Zombie & duplicate jobs
Schedules still feeding deprecated tables, twin pipelines computing the same thing. Turned off, that is pure savings.
Storage layout
Partitioning, clustering, and file sizes that stop full scans: the difference between minutes and seconds on point queries.
Spot & instance strategy
Interruption-tolerant workloads moved to spot capacity and the right instance families for the job.
Warehouse sizing & caching
SQL warehouses sized to real concurrency, with caches actually getting hits.
// Representative: a 40 TB fact table with no clustering — point queries went 12 min → 8 sec after a layout fix.
Who This Is For
A Fit If You're a...
- ✓US regional or super-regional bank ($10B–$100B AUM)
- ✓US PE-backed mid-market firm ($50M–$500M revenue)
- ✓US healthcare payer, provider, or pharma company
- ✓RegTech or fintech firm running production workloads on Databricks or a comparable lakehouse
- ✓Real platform spend ($200K+/year) with no fully-staffed in-house Databricks/Snowflake Center of Excellence
Better Served Elsewhere If...
- —You're pre-Series-B
- —You already have an in-house platform team of 8+ engineers covering this ground
- —Your annual platform spend is under $200K, so the fixed fee won't pencil out yet
Frequently Asked Questions
What happens if the findings are thin? +
That's covered by the guarantee. If the annualized savings we identify don't exceed the audit cost, you don't pay. In practice, every platform we've assessed has had at least one finding that alone justified the fee.
What exactly does "guarantee" mean? +
We calculate annualized savings directly from your cluster telemetry and job run history: actual compute costs, not estimates. If that number, summed across every finding, is less than what you paid for the audit, we refund the fee in full.
Is this covered by an NDA? +
Yes. We sign your NDA or ours before access is provisioned. Findings, telemetry, and any data we touch are treated as confidential and are not referenced in future case studies without separate written permission.
What access do we need to provide? +
Read-only access to your Databricks workspace (or comparable lakehouse), the associated cloud account, and the Git repo containing your pipeline code. No write access is required at any point.
Ready to See What Your Platform Is Costing You?
Start a conversation about the Health Check: fixed scope, fixed fee, guaranteed findings.
Start a Conversation