Bias for Action: Convincing Others to Act

Situation

During a critical holiday shopping season, our site had a checkout performance issue that caused random delays. Engineering suspected a caching issue, but no clear RCA had emerged. Several leads wanted to wait for full diagnostics before making any changes — but I felt strongly that we couldn’t afford inaction during peak traffic.

Task

I took ownership of proposing a safe, short-term fix that could reduce risk immediately, while the long-term analysis continued. My challenge: to convince senior engineers to act quickly without full proof — and take partial steps under uncertainty.

Action

I pulled logs manually and found a pattern: nearly all laggy checkouts were associated with missing cache headers on promo-related assets. I mocked a fix locally, showing 40% faster response times. I documented a rollback plan, shared it in Slack, and invited key engineers to a 15-min sync with a demo.

My proposed patch:

• 📦 Explicit TTL header for promo asset responses
• 🛑 Added fallback cache to CDN edge node
• ⏱️ Rollback script in case of side effects

// Fast fix: cache-control middleware
  app.use('/promo', (req, res, next) => {
    res.set('Cache-Control', 'max-age=600, public');
    next();
  });
      

Result

The team approved the patch. After deployment, median checkout times improved by 32% in just 1 hour. Our exec team noted this improvement during the weekly ops review, and I was invited to draft a "fast-path patch" protocol for future incidents.

What could’ve taken days of cautious delay instead became a same-day improvement — thanks to structured risk-taking and fast prototyping.

Reflection

• ⚡ Bias for action isn’t reckless — it’s about informed urgency.
• 🧭 You can influence without authority by showing proof, not just opinion.
• 🚦 Fast, reversible decisions are often better than slow, perfect ones.

FAQ