Sampling from rare, hidden and hard-to-access populations and creating estimation methods to obtain efficient information from the sampled data are core challenges in sampling theory and applications. This paper builds on Adaptive Web Sampling, proposed by Thompson(2006). These new sampling strategies extend previous adaptive network sampling methods to have more flexibility in controlling sample coverage. Here we propose a new application to investigate financial crimes such as money laundering, where the criminal transforms illegally derived funds into an apparently legal source. Financial institutions are developing anti-money-laundering strategies and training their associates to detect these fraudulent financial activities. We show designs that can efficiently assist in detecting highly risky accounts using wire transactions.