How do you set goals for an early-stage product?

One number per bet, and a validation question: what must we see in ninety days to keep believing it? Goals are how the critical question gets answered. They are not targets to hit; they are thresholds below which the bet is in doubt. Metrics inform the strategy and never set it, which is why each goal has a bet above it and not the other way round.

How do you connect metrics to product strategy?

One number per bet, and a validation question: what must we see in ninety days to keep believing it? Goals are how the critical question gets answered. They are not targets to hit; they are thresholds below which the bet is in doubt. Metrics inform the strategy and never set it, which is why each goal has a bet above it and not the other way round.

What is a good ninety-day goal for a startup?

One number per bet, and a validation question: what must we see in ninety days to keep believing it? Goals are how the critical question gets answered. They are not targets to hit; they are thresholds below which the bet is in doubt. Metrics inform the strategy and never set it, which is why each goal has a bet above it and not the other way round.

Goals · Product Strategy · Product Thinking

§The question: how do we measure progress?

§Each in the strategy carries a critical question. Goals are how the question gets answered: a number, a threshold, and a date. This chapter is short because the discipline is simple and the failure is subtle.

§ 7.1One number per bet#

§Not five. One. The bet says "the customer will build her own profile without an incentive." The critical question asks whether that is true. The goal says: within ninety days of launch, more than half of guests who are shown the profile complete it, with no reward attached. One number, one threshold, one date.

§Andy Grove's original discipline, which John Doerr carried out of Intel and into the OKR movement, was exactly this: an objective with a small number of measurable results, each of which would tell you the objective was being met. At a startup the objective is the bet and there is room for one result. More than one and the team will hit the easy one and call the bet validated.

Weak goal	Strong goal
"Grow user engagement and improve retention."	"Sixty percent of guests shown the profile complete it, unprompted, by day ninety."
"Validate the operations use case."	"In the pilot restaurant, servers open a regular's profile card before seating on at least half of shifts in month two."
"Launch payments and increase revenue."	Not a goal, because there is no bet above it in the sequence yet.

§The third row is the important one. A goal with no bet above it is a metric that has started to drive the strategy instead of informing it. The first whitepaper made the argument that metrics inform and never set; the ladder is the mechanism that enforces it. Every goal has a parent bet. A number nobody can trace upward is a number the team is chasing for its own sake.

§ 7.2The validation question#

§The frame for each goal is a question, not a target: what must we see in ninety days to keep believing this bet? The framing matters because it sets what happens when the number is missed. A missed target means the team failed. A failed validation means the bet is in doubt, and the strategy changes above the roadmap, which is what flexible on the details was for.

§ 7.3Ninety days, and why#

§Long enough for real behavior to show. Short enough that a wrong bet costs one quarter, not a year. And, at a startup, roughly the interval at which the team can honestly reassess without either panicking at noise or coasting on hope. The number is not sacred; a bet about a weekly behavior may resolve in thirty days and a bet about an operation's habit may need a season. But every goal has a date, and the date is when the team reads the number and decides, on the record, whether the bet holds.

§ 7.4What goals are not#

§They are not the North Star metric, which comes later, when the exists and the sixth whitepaper's loops are running. They are not a dashboard. At this level of the ladder there are three to five goals, one per bet, and the team can hold all of them in its head. A team that needs a dashboard to know whether its bets hold has too many bets.

§They are also not a promise to investors. A goal that is reported outward becomes a target, and targets get hit by moving the definition. Keep the goals internal until the bet has held twice.

§ 7.5What comes out#

§One line per bet: the number, the threshold, the date, and the sentence beginning "if this is below the threshold, the bet is in doubt because..." Three to five lines. The next chapter asks what to build to move them.

7. Goals