CTO Leadership Rules

Technical Strategy

• Technology serves business goals — cool tech that doesn't move metrics is a hobby
• Build for current scale, architect for 10x — over-engineering kills startups, under-engineering kills scale-ups
• Buy vs build: build only what's core differentiator — everything else is distraction
• Standardize stack across teams — consistency beats local optimization
• Technical roadmap ties to business roadmap — explain why in business terms

Architecture Decisions

• Document decisions with ADRs (Architecture Decision Records) — future you needs context
• Reversibility matters: prefer decisions you can undo — paint yourself into fewer corners
• Boring technology for critical paths — innovation in one layer, stability in others
• Monolith first, microservices when you feel the pain — premature distribution is expensive
• Every dependency is a liability — evaluate maintenance burden, not just features

Technical Debt

• Debt is intentional trade-off, not accidental mess — if it wasn't a conscious choice, it's just bad code
• Track debt explicitly — backlog items, not mental notes
• Pay interest continuously: 20% time for maintenance — big rewrites fail, steady improvement works
• Refactor alongside feature work — pure refactor sprints lose business support
• Some debt is fine — shipping matters, perfectionism kills companies

Team Building

• Hire for slope, not intercept — growth rate beats current skill for junior roles
• Senior engineers multiply others — evaluate by team output, not just individual contribution
• Specialists and generalists both matter — T-shaped teams with deep experts
• Promote from within when possible — external hires reset culture
• Fire fast when values don't align — skills can be taught, values can't

Engineering Culture

• Blameless postmortems — focus on systems, not individuals
• Code review is teaching, not gatekeeping — explain the why, not just the what
• On-call must be sustainable — burnout kills retention
• Psychological safety enables innovation — people who fear failure don't experiment
• Celebrate shipping, not just building — output matters, not just effort

Scaling the Org

• Small teams (5-8) with clear ownership — bigger teams diffuse responsibility
• Conway's Law is real: org structure becomes system architecture — design both together
• Process scales humans, not replaces judgment — too much process slows everything
• Communication overhead grows O(n²) — add coordination roles before it breaks
• Written culture scales better than meeting culture — decisions in docs, not Slack threads

Metrics

• DORA metrics for team health: deployment frequency, lead time, failure rate, recovery time
• Track cycle time end-to-end — where does work wait?
• Incident count and severity — reliability is a feature
• Technical debt ratio — subjective but discuss it
• Don't measure lines of code or commit count — gaming metrics destroys culture

Business Interface

• Translate technical risk to business risk — "this could cause 4 hours downtime" not "race condition"
• Give options with trade-offs, not just recommendations — empower decisions
• Say no with alternatives — "we can't do X, but we could do Y"
• Protect the team from thrash — absorb priority changes, don't relay every pivot
• Revenue, retention, and reliability — know how your systems affect each

Staying Technical

• Code occasionally, but don't block critical path — review PRs, build internal tools, prototype
• Stay in on-call rotation at reduced frequency — feel the pain your team feels
• Architecture reviews keep you connected — where decisions have most leverage
• Learn new tech on side projects — production isn't for experimenting
• Your job shifts from doing to enabling — accept the transition

Common Mistakes

• Hiring too senior too early — expensive people need leverage that doesn't exist yet
• Building platform before product — internal tools are cost centers until product-market fit
• Rewriting instead of refactoring — big bang rewrites take twice as long and often fail
• Ignoring security until breach — security is cheaper proactively
• Letting tech debt become a blocking crisis — steady payments beat bankruptcy