Skip to content

Feature Flag Limitation Reduces System Volatility

by Matt McGinnis on December 28, 2025

Situation

  • Rippling, a rapidly growing business software platform valued at over $16 billion, has a large R&D organization with 1,300 people
  • The product team had established a "Product Quality List" (nicknamed "the pickle") as a lightweight checklist to ensure consistent quality standards across products
  • During a critical product release, CEO Parker Conrad attempted to install a new feedback application and encountered a blank screen
  • The issue was traced to a forgotten feature flag that hadn't been disabled before release
  • This represented a serious quality control failure that could impact customer experience

Actions

  • Matt McGinnis (CPO) immediately addressed the issue with the team, providing direct feedback about the mistake
  • Rather than just fixing the immediate problem, McGinnis questioned how it was missed in their "factory inspection process"
  • He identified a systemic gap: their quality checklist had no specific guidance for feature flag management
  • McGinnis added a new standard to the pickle: "You are allowed to have one feature flag that governs your entire product at ship"
  • This was deliberately set as an extreme standard that might not always be achievable but established a clear aspiration

Results

  • The team now had a clear guideline for feature flag management during releases
  • The pickle continued to evolve as a "lightweight way to lower the beta of the system" (reduce volatility)
  • The approach maintained a balance between quality control and innovation speed
  • The process improvement was implemented without creating excessive bureaucracy
  • The team could continue to iterate quickly while avoiding this specific failure mode

Key Lessons

  • Evolve quality standards through real failures: Use actual incidents to identify gaps in quality processes rather than trying to anticipate every possible issue upfront.

  • Balance process with speed: Create lightweight quality controls that reduce volatility without suppressing innovation. As McGinnis put it, "processes exist for the sole purpose of lowering beta."

  • Set aspirational standards: Sometimes an extreme standard that isn't always achievable ("one feature flag maximum") creates clearer guidance than a complex but more realistic rule.

  • Create memorable frameworks: The "pickle" nickname made the quality checklist more distinctive and memorable, helping it become part of the team's common language.

  • Treat each failure as a system improvement opportunity: Rather than just fixing the immediate issue, use each problem to improve the underlying system that builds the product.

  • Maintain public accountability: McGinnis provided feedback in public channels so other teams could learn from the experience, creating a culture of continuous improvement.