Exaggeration of performance metrics (revenue, product efficacy, ad viewership, etc.) by entrepreneurs to investors and clients is a common problem in high tech entrepreneurship. We model this as a principal-agent problem in a contract-theoretic setting, where the entrepreneur (agent) can undertake costly actions to strategically lie about a key performance metric (agent type) in order to extract higher payment from the investor (principal). We demonstrate that the optimal contract features widespread exaggeration by all entrepreneur types, and the investor exploits it as
a screening mechanism to ordinally rank the entrepreneur by his true underlying type. We study the effect of an audit in which, if caught cheating, the agent pays a penalty. We show that rather than deterring fraud, audits actually amplify the degree of exaggeration.