We study multi-period salesforce incentive contracting when sales agents can dynamically choose between a safe action and a bold or risky action, where the bold action has higher expected sales but also higher polarization in sales and higher effort cost. Using a two-period model with ex-ante independent time periods, we show that a two-period contract is (weakly) optimal and that the firm determines the optimal contract by trading off the number of bold actions taken with the expected compensation per bold action. Under different conditions, the optimal contract is an “account-balance” contract that ignores failure and pays according to the number of periods with a high sales outcome (i.e., it simply gives a reward for every successful outcome), an “extreme” contract that penalizes failure and pays only if there is a high sales outcome in every period (i.e., it sets a very high bar for reward), or an “early-polarization” contract that rewards early failure only if the early sales outcome is either high or low (but not intermediate) followed by a high sales outcome. Specifically, in the early-polarization contract, a reward may be given in spite of early failure purely to make effort levels inter-temporally dependent to reduce the overall cost of motivating effort; this adds to existing explanations, based on career concerns and learning, for tolerating failure. Overall, for moderate levels of increased return from the bold action relative to the safe action, the account-balance contract is optimal, otherwise one of the extreme contract and the early-polarization contract is optimal
