I track the accumulated weight of every field, rule, and object that was added and never removed, because the org gets slower and more fragile with each one. The failure mode is a system so encrusted with orphaned fields and dormant automations that no one can change anything without breaking something they did not know was connected.
Technical-debt metrics are org-specific by nature; the value is the trend, not the absolute number. Anchor on whether debt is being paid down each quarter, not on hitting a universal threshold.
6 indicators
Share of custom fields not referenced by any report, page layout, automation, or integration.
unreferenced custom fields / total custom fieldsBenchmarkPractitioner finding: 20-40% of custom fields on a mature org are unused; target ongoing pruning to keep it under ~15%
Orphaned fields bloat page layouts, confuse reps into filling dead fields, and make every schema change a guessing game about what is safe to touch.
Share of automations that are active but have not fired (or serve no live process) in a long window.
dormant active automations / total active automationsBenchmarkPractitioner target near zero standing; audit quarterly. A rule that has not fired in a year is either dead or a latent landmine
Dormant automations are latent risk: they cost nothing until a data load or a reused value wakes them up and they corrupt records nobody was watching.
Share of fields on a page layout that carry a value on a meaningful portion of records.
fields populated on >X% of records / fields on layoutBenchmarkPractitioner rule of thumb: a field under ~5% populated is a candidate for removal unless it has a specific low-volume purpose
Low-utilization fields are the visible face of debt to reps; they lengthen every page and erode the discipline of keeping any field current.
Count of automations, formulas, or code with hard-coded IDs, usernames, or picklist values that break on change.
BenchmarkPractitioner target as low as possible; each hard-coded value is a break waiting for the referenced record to change or deactivate
Hard-coded references are brittle joints: they work until the underlying record changes, then fail in ways that are slow to diagnose because the value is buried in logic.
Count of components on retired technology (process builders, workflow rules, old API versions, legacy Aura where LWC is the standard).
BenchmarkSalesforce has retired Workflow Rules and Process Builder for new creation; practitioner target is a migration plan with a shrinking count each quarter
Deprecated components accumulate risk as the platform moves on; the count and its trend show whether debt is being paid down or ignored.
Share of deployments or config changes that cause a defect or require a rollback.
changes causing an incident / total changesBenchmarkDevOps practitioner range: elite teams sit under 15%; a high rate in a CRM org usually traces to untested automation tangle
Change failure rate is the bottom-line symptom of debt: the more tangled the org, the more often a safe-looking change breaks a hidden dependency.