How do we assess a week when the main result was learning?
Sejal Amin talked about asking what people learned, which gives a different lens from asking only what they shipped. A learning review still needs to show what changed in the team's understanding, so ask for the question, the evidence and the decision it affected.
A failed assumption can count as learning if the team can explain what it will do differently. Activity alone isn't learning, though, so I'd keep asking what changed. A result might narrow the question without settling it. When that happens, describe the next check and why it's worth doing.
Keep exploratory work separate from delivery commitments. If an experiment was meant to answer a question, look at whether it did. If a customer deliverable was due, don't relabel an incomplete task as learning to dodge the commitment.
The review should also move priorities. Learning matters when it changes a choice, stops unnecessary work or points to a better test. And watch for a trap: swapping a shipping count for a learning count rewards people for producing impressive sounding observations every Friday.