[edmondlau]

Your observation is a key reason why taking the time to define your impact and to measure it is so important. As Drucker says, what gets measured gets improved.

In business contexts, your engineering impact generally ties back to the business value you create (i.e. revenue and profit). If there is already some model for translating your area of work to revenue (e.g. increase users -> increase revenue, reduce fraud -> increase revenue) then that can also give another proxy metric to optimize for. So for example, when working on Google Search Quality, we would often just optimize for long clickthrough rates, knowing that strategically, the ads team would take care of turning returning searchers to revenue.

It gets harder when you're working on areas like bigger bets (where you can have high impact but it is unknown for a long time) or in trying to understand an infrastructure investment in terms of its business value to the company. There, it may be sufficient to just know that something is of strategic value and then measure your impact in terms of those strategic goals.