For teams building a genuine second product (as opposed to Features/Growth/Scaling work on the core), Casey Winters argues OKR-style success metrics are the wrong evaluation tool, because real evidence of "winning" may not exist for a year or more. Instead he proposes a learning feedback loop: a ranked list of assumptions to test, the expected path if those assumptions hold, and a running account of what's actually being learned — fed back to leadership on a fast cadence. This substitutes "are we learning the right things at the right pace" for "did we hit the number," which matters because the PMF bar for a new product can take 1-3 years to clear (see S-Curve Evaluation (Next Growth Wave vs. Enhancer)). Related failure modes when this discipline breaks down: Three Failure Modes for New Products Despite Good Process.
The concrete mechanics of the process: instead of conventional OKR-style metrics (MAUs, revenue), a new-product team maintains a ranked list of assumptions still to be tested, an expected path forward, and explicit statements of what is actually being learned — reported back to leadership on a fast cadence.
The reason for the substitution: real learning and progress on a new product can precede any measurable "win" by as much as a year, so metrics that require a win to register (like OKRs) will read as failure long before the team has actually failed.