Earlier quoted context omitted.
This is why KPIs or targets should NEVER be calculated values like averages or ratios. The team is then incentivized to do something hostile such as not promote the content as much so that the ratio is higher, as soon as they barely scrape past the impressions mark.
Sure they can, e.g. on-time delivery (or even better shipments missing the promised delivery date) is a ratio. Or inventory turn rates, there you actually want people to attack the denominator. Generaly speaking, an easy solution is to attach another target to either the nominator or denominator, a target that requires people to move that in value in acertqin direction. That might even be a different team thanthe one…
These are good in that they’re directly aligned with business outcomes but you still need sensible judgement in the loop. For example, say there’s an ice storm or heat wave which affects delivery times for a large region – you need someone smart enough to recognize that and not robotically punish people for failing to hit a now-unrealistic goal, or you’re going to see things like people marking orders as canceled or faking deliveries to avoid penalties or losing bonuses.
One example I saw at a large old school vendor was having performance measured directly by units delivered, which might seem reasonable since it’s totally aligned with the company’s interests, except that they were hit by a delay on new CPUs and so most of their customers were waiting for the latest product. Some sales people were penalized and left, and the cagier ones played games having their best clients order the old stuff, never unpack it, and return it on the first day of the next quarter - they got the max internal discount for their troubles so that circus cost way more money than doing nothing would have, but that number was law and none of the senior managers were willing to provide nuance.