You work on a B2B product with a low-volume SMB acquisition funnel and want to test a change to an early funnel step. Because traffic is limited, you may need to run the experiment for a long time or accept that only larger effects are detectable. The concern is whether the test can be designed to produce a decision you can trust.
What sample size considerations matter most when you are testing a change in a low-volume SMB funnel? How would you choose an MDE, define the key metric and guardrails, and decide whether the experiment is feasible at all?