Tagging hygiene: cost allocation that holds up
Untagged spend is unowned spend. A practical way to get from "60% allocated" to a bill every team recognises as theirs.
Every cloud cost programme eventually hits the same wall: you can see the total, but you cannot say who caused it. The fix is boring and it is tagging.
Pick five tags, not fifteen
Owner, environment, service, cost-centre, and product. More than that and compliance collapses. Fewer and you cannot answer basic questions.
Enforce at creation time
Retro-tagging is endless work. Enforce required tags in infrastructure-as-code and with policy checks, so untagged resources simply cannot be created in production accounts.
Handle what cannot be tagged
Shared clusters, data transfer, and support fees resist tagging. Allocate them with a documented, boring rule — usage share or headcount — and publish the rule so nobody relitigates it monthly.
Measure allocation coverage as a KPI
Report the percentage of spend mapped to an owner every month. Moving from 60% to 95% coverage changes the conversation from "cloud is expensive" to "this team's pipeline doubled".
The payoff
Once allocation holds, showback becomes possible, unit economics become real, and savings work goes to the teams that can actually act on it.