Response times went from 800ms to 1.2 seconds after we added Redis. We made it slower.

What Went Wrong

We cached full API responses including user-specific data. User A's dashboard got cached. User B got User A's data from cache, then made extra queries to fill gaps. More queries = slower.

The Security Near-Miss

Cache keys were URL-based, not auth-based. We caught the performance issue before user data leaked across accounts—but barely.

How We Fixed It

• Cache expensive, non-personalized queries only
• Include user IDs in cache keys for user-specific data
• Use TTL-based expiration with manual invalidation for critical updates

const cacheKey = `user:${userId}:permissions`;

Response times dropped to 240ms. Database load down 60%.