Platform audit and cost management

Most data platforms carry meaningful waste: compute nobody right-sized, storage nobody tiered, pipelines nobody retired, and contracts nobody renegotiated. I find it, quantify it, and remove it. This is the most measurable work I do, and the numbers below are why clients usually start here.

What’s usually broken

  • Nobody can say what a query costs. Spend is one line on a cloud bill, with no attribution to teams, pipelines, or use cases.
  • The pricing model predates the workload. On-demand where reservations would win, or reservations sized for a peak that no longer exists.
  • Storage grows forever. No lifecycle policies, no partitioning discipline, snapshots kept out of habit.
  • Dead work runs nightly. Pipelines and scheduled queries feeding dashboards nobody has opened in a year.
  • Vendor spend rides inertia. Seven-figure contracts renewed annually because migration feels scary and nobody has done the analysis.

What I do

  • Full platform audit: a fixed-scope review of your data estate covering cost, architecture, quality, security, and compliance. You get ranked findings with the money and effort attached to each, and a roadmap ordered by payback.
  • BigQuery cost work: slot reservation and on-demand mix analysis from your actual job history, partitioning and clustering rationalization, storage lifecycle policy, and dead-query elimination.
  • Cost attribution and chargeback: per-team, per-pipeline, per-dataset showback built from labels plus attribution logic for the shared projects where labels fail. FinOps dashboards and budget alerts that reach the responsible team.
  • Vendor and contract analysis: build-versus-buy and migration analysis with the full cost of both paths, written down.
  • Second opinions: a stalled or expensive implementation left by a previous partner, reviewed by someone with no stake in the original decision.

How it works

Audits are fixed scope and short, typically a few weeks: one phase of analysis on your job history and billing data, one phase of findings, ranking, and no-regret fixes. The deliverable is yours either way: a findings report and roadmap your team can execute without me.

Proof

At TelevisaUnivision I cut GCP spend by more than $1MM a year through the BigQuery and storage work described above, built the cost-attribution model that kept it cut, and led the analysis and execution of a LiveRamp exit worth over $8MM a year. At AXS I designed a Snowflake spend reduction projected at over $500K a year. Career total: eight figures of annualized savings.

Start a conversation

Bring 90 days of billing history and I’ll tell you quickly whether an audit is worth your money.

sandro@engramdataworks.com