What is being tested
Did tag combinations create the billed unique metric series under the documented rule? The boundary for this investigation is monitoring custom-metric cardinality billing. Begin with the disputed transaction or population, then identify which metric series inventory establishes the observed position and which tag configuration history supports the comparison. A difference in totals should not replace this question.
Evidence: metric series inventory
For monitoring custom-metric cardinality billing, metric series inventory must be linked to tag configuration history. Retain stable identifiers and their effective relationships. Current labels are insufficient when assets or accounts changed during the period. Explain one-to-many relationships explicitly and preserve the history needed to distinguish an alias, replacement or reassignment from a genuinely additional item.
Evidence: tag configuration history
For monitoring custom-metric cardinality billing, tag configuration history must be linked to series-count definition. Document the observation window, units, inclusion criteria and export version. Identify gaps and corrected events before using the total. Keep raw observations separate from derived quantities so a reviewer can reproduce the population without assuming every logged event is independently chargeable.
Evidence: series-count definition
For monitoring custom-metric cardinality billing, series-count definition must be linked to monitoring invoice. Retain the applicable wording, effective dates and scope of covered transactions. Identify the event or population that controls the calculation. Do not silently replace a contractual definition with a dashboard label, customary practice or the latest published rule.
Evidence: monitoring invoice
For monitoring custom-metric cardinality billing, monitoring invoice must be linked to metric series inventory. Keep the issued document version and line-level quantity, currency and service period. A header total cannot establish which component is being tested. Retain later corrections as linked versions, so a replacement does not create a second liability.
Reconciliation logic
Reconstruct unique series over the measurement window and identify tag-driven growth separately from duplicated meter counts. Build the comparison at the level identified by metric series inventory and retain the governing version from tag configuration history. Show intermediate classifications and excluded items separately; a net total can hide an unsupported component or a correctly offset correction.
Exception conditions
High cardinality may be valid consumption from a configuration change. Treat the item as an unresolved exception only when the comparison described here cannot be supported by the linked metric series inventory, tag configuration history, series-count definition, monitoring invoice. Document the conflicting input or rule. A plausible operational explanation requires validation, but it should not be discarded to maximize an apparent financial difference.
Human review and outcome
Observability engineering validates series lifetime and aggregation. Classify quantity disagreement separately from an approved cardinality reduction. Keep the reviewer's reason and source references with that disposition. A supported correction should be followed to the revised record or settlement; an accepted explanation can close the question with no adjustment. Missing authority or evidence should remain an open task rather than a confirmed recovery.
Limitations and processing boundary
Do not infer license removability from activity alone or describe a proposed configuration change as confirmed savings. The authoritative spend, license and contract producer is not complete; customer evidence and processing validation are prerequisites to production conclusions. In this scenario, absence of metric series inventory or tag configuration history limits whether the comparison can be completed. The review method describes what people should validate, not a promise that AuditRes automatically detects or executes this specific outcome.
AuditRes pathway
Discuss monitoring custom-metric cardinality billing in the Technology Spend workspace. Review current plans, the shared platform and secure evidence requirements; use the existing contact path to confirm the sources and validation this scope requires.
AuditRes Technology Spend: Available for onboarding. Public previews use synthetic demonstration data; production processing remains gated until applicable customer sources and authoritative processors are connected and validated.
Neighboring financial questions
- Review billed API calls across pagination and export requests
- AI cached-input price classification
- AI embedding refresh versus incremental indexing costs
- AI asynchronous job cancellation billing boundaries
Technology Spend resource hub · All guides in this evidence collection