Start with the channel at the intersection of high spend and high uncertainty, such as retargeting or branded search, where over-credit is most likely (see Deterministic (in-platform) attribution).
Tests can also show that upper-funnel channels (connected TV, YouTube, podcasts, out-of-home) are under-credited. The goal is causal truth, which sometimes means spending more. Related: Unmeasured channel budget allocation, Blended-efficiency evaluation of unmeasured channels.
Pick the first channel to test as the one that looks great in attribution dashboards but whose incrementality is doubtful, such as retargeting or branded search.
A channel that is good at finding people already deep in the purchase funnel is exactly where attribution over-credits.
Also test upper-funnel channels such as connected TV, YouTube, podcasts or out-of-home. They influence purchases days or weeks later through a different channel, so last-click style credit goes to the channel that gets the clicks. That leaves them under-credited. Geo tests (geographic-experimentation) suit these channels. As the source puts it: "Incrementality testing isn't just about identifying waste, it's about getting closer to causal truth."