Cycle counts die when they are a yearly panic. They work when they are a daily trickle of the bins most likely to be wrong.
I pull a queue from SQL: high-velocity SKUs, bins with old last-counts, locations that had a mis-pick last week. The phone shows one bin at a time. Count, submit, next.
Do not fight the WMS if you cannot replace it
Write results to a table you own, then export in the format the WMS already accepts. A night job that posts adjustments is less risky than living in two systems that disagree all day.
Exceptions — cannot find the bin, cannot identify the item — should become visual-lookup jobs, not skipped rows.
Show the delta
People work harder when they can see variance dropping. A tiny dashboard of counted vs remaining beats a motivational poster.
