Dual-track discovery
Dual-track discovery runs a continuous learning track alongside delivery, so the work of finding out what is worth building happens outside the OKR set rather than being graded as a goal. Discovery graded as a goal stops being discovery, because nobody reports an answer that makes their number look bad.
Why grading discovery breaks it
A discovery track exists to find out whether something is true. Its most valuable output is often "no", and a "no" that costs somebody a grade will be reported as "promising, needs more time".
So the moment you put discovery in the goal set with a target attached, you have created a system that punishes the answer you most need. It will keep running, and it will stop telling you anything.
What the two tracks do
The delivery track builds and ships against outcomes you have decided to pursue. It carries the OKRs and the confidence checks.
The discovery track runs continuously underneath, testing assumptions and feeding a maintained list of what might be worth doing next. Its cadence is not the quarter, and its output is decisions rather than deliverables.
How the two connect
Through the ideas list. Discovery produces candidates with evidence attached; planning pulls from that list rather than from whatever was said loudest in the room.
The connection point is where most companies lose it. Run discovery with nowhere for its findings to land and you have a research function that nobody reads, which is a common and expensive shape.
What is dual-track discovery? +
Running a continuous discovery track alongside delivery, so learning work happens outside the OKR set instead of being graded as a goal.
Should discovery work be an OKR? +
Generally no. Discovery’s most useful output is often a negative answer, and grading it makes reporting that answer costly. Keep it continuous, and let it feed the ideas list that planning pulls from.