▶ YouTube Summaries

• YT :: https://www.youtube.com/watch?v=6tRjKSwL0Qk
• Original title :: Want a Promotion? Ask for a Test

Karell argues that a CTO title, a bigger team or a finished course do not prove that someone can make the decisions the role exists for — whether the company should fund a solution at all. More implementation scope makes an engineer better at implementation, not at consequential technology choices. The fix is not to grab authority but to ask an authorized sponsor for a bounded build-vs-buy or investment assignment, with explicit limits and an outcome review, producing an inspectable record of decision judgment.

Start with the mandate, not the title

• "CTO" means different jobs: product technology bets, engineering scale, internal systems, security — often tangled together.
• Find out what the target company needs its CTO to decide, protect, fund or stop, then pick the stretch assignment to match.
• Mandate = engineering scale → people leadership and operating systems may be the missing work.
• Mandate includes product bets, capital allocation, technology risk → a consequential choice in that territory is needed.
• Technical credibility, management ladders and courses still matter when they add needed responsibility.

The fictional decision-gap model

• Explicitly not a readiness score or career test — an illustration.
• Baseline: sponsor already picked the implementation; the candidate builds it, accepted at time 2. Legitimate work, but no option comparison, no bounded choice, no decision artifact.
• Corrected route: candidate analyses the choice → sponsor reviews → candidate gets a narrow right to run the recommendation or pilot → outcome review. Sponsor keeps approval and residual risk. Finishes at time 6 with one decision-and-review artifact.
• Trade-offs:
• The route is slower and costlier — too much process for a tiny, clear decision.
• Remove the sponsor and the route stops after analysis with a recommendation waiting. That shows access matters, not that the candidate lacks judgment — authority that was never granted cannot be demonstrated.

What a real decision assignment includes

• One target mandate and one capability gap; a concrete build-vs-buy or investment decision, not "a chance to be strategic".
• Agree up front on: business outcome, the decision or recommendation owned, spending and risk limits, approvals the sponsor keeps, people who provide context, review date.
• Compare plausible options: engineering work beside total cost, integration, exit cost, staffing, risk, uncertainty, and the strongest competing use of the same resources.
• State what would make the recommendation wrong — it makes the trade-off inspectable.
• At the outcome review, compare result vs expectation and separate own contribution, team contribution and uncontrolled conditions.

Takeaway

• One assignment does not certify anyone for every CTO job, but it is real evidence for the next development conversation — a record of decision judgment inside authority someone actually gave.