We outgrew homegrown feature flags

That usually means it’s time to treat feature flags as a product, not a side utility.

A good next step is to decide whether you want to:

1. Buy a flagging platform

   • Best if you need:
     • targeting/audiences
     • gradual rollouts
     • kill switches
     • experiments
     • audit logs / approvals
     • SDKs across services/languages
     • reliability and low-latency evaluation
   • Common options: LaunchDarkly, Unleash, ConfigCat, Statsig, Split, Flagsmith

2. Keep it in-house, but formalize it

   • Best if:
     • your use cases are simple
     • you have strong platform engineering capacity
     • compliance/data residency is critical
   • You’d want:
     • a proper flag service
     • admin UI
     • RBAC
     • flag lifecycle management
     • deletion/expiry policies
     • analytics/audit trail
     • SDKs and caching strategy

3. Hybrid approach

   • Buy for core runtime evaluation, keep internal tooling for workflows, metadata, and governance.

Signs you’ve outgrown homegrown flags

• Developers don’t trust them in production
• No clear ownership or expiration policy
• Flags accumulate and never die
• Rollouts are hard to target or automate
• Evaluation logic is duplicated across services
• There’s no auditability or access control
• Incidents require manual coordination to flip flags fast

Migration approach

• Inventory existing flags
• Classify them:
  • release flags
  • ops flags
  • experiment flags
  • permission flags
  • config flags
• Identify the highest-risk flags first
• Run both systems in parallel
• Add expiration and cleanup automation
• Move consumers incrementally by service or flag type

Key questions to choose a path

• How many flags do you actively use?
• Do you need per-user targeting?
• Do flags gate experiments or only releases?
• How many services/languages consume them?
• What’s your uptime/latency requirement?
• Do you need approvals, audit logs, or SOX/compliance support?

If you want, I can help you with either:

• a buy vs build decision framework,
• a migration plan off homegrown flags, or
• a requirements checklist for evaluating vendors.