Reporting and Showback | DevOps Peach

Good FinOps reports are built per persona — trend and unit cost for leaders, budget versus actual for finance, daily service cost for engineers — and showback makes each team see its own spend without moving money.

A report is useful only if someone changes a decision because of it. Start from the question, not from the data.

flowchart LR
  D["Allocated cost data"] --> L["Leadership: trend, unit cost"]
  D --> F["Finance: budget vs actual vs forecast"]
  D --> E["Engineering: daily cost per service"]
  D --> P["Product: cost per customer or feature"]
  E --> ACT["Action: fix, schedule, rightsize"]
  %% step D: All reports read the same allocated, normalised data so numbers match.
  %% step L: Leaders need direction: total trend, unit cost and top risks, monthly.
  %% step F: Finance compares budget, actual and forecast by cost centre.
  %% step E: Engineers need near real-time cost per service, ideally in their own dashboards.
  %% step P: Product looks at cost per customer, plan or feature.
  %% step ACT: The goal of every view is an action, not a nicer chart.

Report building blocks

View	Shows	Cadence
Executive summary	Total vs budget, month-over-month, unit cost, top 5 movers	Monthly
Team showback	Each team's effective cost, trend, top services, untagged share	Weekly
Engineering daily	Cost per service and environment, yesterday vs 7-day average	Daily
Optimization tracker	Open recommendations, savings realised, savings missed	Weekly

Showback done well

Use EffectiveCost (amortized, after discounts) so teams are not punished for the month a commitment was bought.
Show trend and context, not just a number: "+18% because traffic grew 25%" is good news.
Include the untagged and shared lines so totals add up to the invoice.
Rank the top movers — what changed most since last week — because that is where questions start.

Tools

Cloud-native: AWS Cost Explorer and Cost and Usage Dashboards, Azure Cost Management, GCP billing reports and Looker Studio. Across clouds: FOCUS data in a warehouse with any BI tool, or FinOps platforms. The tool matters less than one source of truth that every report reads.

Traps: different reports showing different totals (pick one cost metric and document it); dashboards with fifty charts nobody reads; and percentages without absolute numbers — a 300% increase on $5 is not an incident.