Cost Allocation & Showback Platform  ·  View 04 of 22  ·  2 · People and journeys

Journey — Explain a Cost Spike

The load-bearing journey. The trough is 'Explain', and most of the architecture exists to make that phase survivable.

Editable source SVG draw.io All views
Engineering lead owns 9 services Goal — Find out what moved, why, and whether it was us — before the review meeting Trigger — An anomaly alert, or a finance question about last month's number Done when — A one-sentence explanation naming a resource, a day and a cause 1 · Notice alert or question 2 · Locate which dimension moved 3 · Explain ◆ moment of truth 4 · Act 5 · Defend at the review ◆ moment of truth What they do Opens the alert Filters to the delta Compares to baseline Drills to resources Reads the rule that fired Fixes tags or workload Sets a budget alert Shows the statement What the platform gives Alert names the service Attributed dimension Drill to cost records Apportionment maths Coverage reconciler Frozen statement v1 Input versions shown How it feels Confident Uneasy Exposed Where it hurts Alert 3 days late Spike is in 'untagged' Shared cluster: no answer Number changed since Finance has a different figure What must be true Detect within 24 h Remainder under 2% Drill-through p95 10 s Input versions on view Owner-facing gap list One frozen number Journey — An Engineering Lead Explains a Cost Spike The trough is 'Explain'. Every structural decision in this set exists to make that phase survivable. v 1.0 · owner Platform Architecture · date 2026-09

What the trough actually is

  • The spike lands in 'untagged' and the lead cannot tell whose it is
  • The spend is inside a shared cluster and the platform has no per-workload answer
  • The number has changed since the lead last looked, and nothing says why

Structural requirements this journey creates

  • Drill-through from any figure to cost records, p95 ≤ 10 s — a requirement of the ledger, not the dashboard
  • Container-level apportionment, because 'the cluster' is not an owner
  • The three input versions printed on every figure, so 'why did it change' has a one-sentence answer
  • Unallocated remainder ≤ 2%, published per team in currency

Where the journey ends badly

  • Finance shows a different number in the same meeting — two paths, one truth is the failure mode act 5 exists to prevent
◀  Actors and Their Core Journeys All views Journey — Close and Publish the Month  ▶
Solution Architecture v1.0 · Microsoft Azure · Cost & Efficiency Architecture · 2026-0904-journey-explain-a-spike