Rewarding Decisions vs Outcomes

I had a very interesting discussion with Cam MacRae on AVC about how we incentivize staff, especially executives.

One of the key - if not the key - lesson I learned in decision sciences, and apply regularly in business life, is that there is a distinction between good decisions and good outcomes.

• Good decision: a decision you make, given limited information at the time of the decision, that has the best expectation (defined mathematically or any other way you like) of outcome.
• Good outcome: the results you want (increased market share, profitability, revenues, etc.)

Never confused the two. As managers, we can only control our decisions. We have a fiduciary responsibility to make the best decisions we can make, i.e. to maximize the probability of getting the outcome we desire. But there is no guarantee that even the best decision will have a good outcome.

For example, if you have two courses of action: one has a 40% chance of increasing your profitability and one a 70% chance, then a good decision is to take the course of action with the 70% chance. However, even if you take that decision, there is still a 30% chance that you might have no increase in profitability.

As executives, we incentivize people based on outcomes, not decisions. The reasons for this are two fold:

1. Outcomes are easy to measure and agree upon. The CMO and Board both agree on the results and the reward if the incentive plan is $10,000 for each 1% of market share in the calendar year.
2. People can be forgiving of bad outcomes from good decisions; markets are not, and the staff are tied to the company's fortunes. Whether we made a good decision or not, if the company does poorly, there is simply less cash to go around.

Cam said that we currently reward people as follows:

• Good decision + good outcome: bonus
• Good decision + bad outcome: smack
• Bad decision + good outcome: bonus
• Bad decision + bad outcome: smack

His argument is that we really should be rewarding on decision and outcome, not one or the other alone. So the incentive structure should be more along the lines of:

• Good decision+ good outcome: ticker tape parade
• Good decision + bad outcome: nada (no bonus, but not smack either)
• Bad decision + good outcome: smack
• Bad decision + bad outcome: smack

In other words, if you made a good decision with a bad outcome, we are neutral; bad decisions get smacked, even if they succeed against the odds; and good decisions with good outcomes get celebrated.

This is a very interesting thought process, a newer way to compensate. I have two issues with it:

• Measuring decisions is extremely hard. The higher up you go on the ladder, the more important it is that the measures be extremely clear. So while rewarding outcomes is partially because the manager is tied to the fortunes of the company, it is also because outcomes serve as the best measurable proxy for decisions people could think of.
• Outcomes actually are not too bad a proxy for decisions over time. After all, if a good decisions is defined as one which has the best expectation for outcome for the company, then, over multiple decisions, good decisions should lead to good outcomes, and vice-versa.

I think, then, that remaining with the outcomes measurement actually is the best (and cleanest) way to incentivize managers and executives. However, we need to ensure that these are over time, so we do not confuse a bad outcome (which may come from a good or bad decision) with a preponderance of bad outcomes (which should come from bad decisions) and vice-versa. The issue is not measuring decisions vs outcomes; it is measuring one outcome vs multiple

• One good outcome - small reward
• One bad outcome - nada
• Track record of good outcomes - ticker tape parade
• Track record of bad outcomes - smack