What your company is actually optimising for
Lesson 12 of 24, level 01Index
Two teams are both certain they are doing the right thing, and their work directly cancels out. Nobody is wrong. Nobody has read the same goal.
North star, OKR and health metric are three different things
A goal has three different jobs
- North star: Lasting user value
- Quarterly OKR: Change this quarter
- Health metric: Protect this baseline
The north star is the one number that means the product is working. OKRs are what you will move this quarter. Health metrics are the ones you never want to move at all.
Confusing them is the most common metrics mistake in product, and it is what produces two teams whose work cancels out.
A team can agree that a feature is useful and still choose not to build it. Your proposal needs to explain which current outcome it supports and what it would displace.
Use your chosen course project throughout. The additional examples below are fictional practice cases; transfer the method to your own evidence.
Translate a goal into behavior
“Grow” is not enough to guide a choice. Ask which customer behavior needs to change, for which population, and by when. Distinguish a measurable outcome from a deliverable: launching a dashboard is work completed; helping more eligible users finish a valuable task is a change in behavior.
Find the limiting step
Trace how your team's work could influence the outcome. If the current constraint is unreliable data import, polishing a report may not help users reach value. This is a hypothesis about the system, so write down what evidence would contradict it. Team ownership does not automatically mean team causality.
Name the sacrifice
A strategy matters when it changes a choice. Write what the team will delay or decline while pursuing this outcome. Include a guardrail so a local win does not cause a larger loss. A shorter signup flow that creates more unusable accounts may improve one chart while damaging the actual experience.
Make the goal actionable
- Outcome: Whose behavior?
- Constraint: What blocks it?
- Bet: What will change?
- Guardrail: What must not worsen?
Doing it with AI, and where it breaks
Give a model your company's stated goals and ask where two of them would conflict in practice, with a concrete example of a decision that would satisfy one and damage the other.
It will find polite theoretical tensions. Push for a specific decision on a specific screen, otherwise it is a management platitude rather than a finding.
Your AI workbench
Start with your own notes or clearly labelled practice data. Remove private details before sharing. Replace the placeholders, run the prompt in your chosen AI tool, and keep the output beside its source.
Here are the stated goals, evidence, and proposed work [paste]. Separate outcomes from outputs. Map each proposal to a behavior change and identify unsupported causal links. Suggest one guardrail and one item to defer. Ask for missing priorities instead of inventing leadership intent.
Before you use the output
- The outcome names a population and window
- The proposal addresses a plausible constraint
- The deferred work and guardrail are explicit
Stuck? Try this next
When AI produces a generic strategy statement, require it to compare two actual backlog items and explain what new evidence would reverse its recommendation.
Keep a brief AI log: input used, useful output, what you checked, and what you rejected. The decision remains yours.
Build it
A one page map of your company's real goal hierarchy, with one conflict named.
Checkpoint
If you did the build, these take two minutes. If you cannot answer one of them, that is the part to go back to.
Moving on marks this lesson complete. Finish the build first, it is the part that counts.