Field coverage & gaps

An importable record is not the same as a useful one. Every CI class has fields its downstream consumers need — and every discovery source covers only a subset of them. Field coverage is where estates quietly fail: the devices are all there, and half the fields that make them operable are empty.

Coverage, measured per class

ServiceMatch measures field coverage against the target schema, class by class: which required and recommended fields are populated, which are missing, and for missing ones — which records. Rolled-up class rows drill down to the exact devices missing each field, so "serial coverage is weak on network gear" becomes a working list, not a sentiment.

Gaps traced to sources

Because every populated field carries per-field source provenance, gaps trace backward: a missing field is either a source that carries it but wasn't mapped, a source that should be onboarded, or a field genuinely absent from the estate's tooling. Those are three different fixes, and the analysis tells you which one you have.

The improvement loop

• →Evidence recommendationsquantify what a candidate fix is worth in device counts — "mapping this field lifts these N records" — never in projected percentages.
• →Foundation gaps are reported separately: missing users, departments, and locations undermine everything modeled above them, so they get their own report rather than hiding inside device metrics.
• →Export quality gates close the loop: bundles that fail required-field checks are blocked, not shipped with holes (see Policies).