Decisions are often delegated to experts chosen based on their past performance record which may be subject to noise. For instance, a person with little skill could still make a lucky decision that proves correct ex-post, while a skilled expert could make the best possible use of available information to reach a decision that, with hindsight, turns out incorrect. We aimed to show that one could assess decision skills more accurately when analyzing not only the observed decisions, but also the decision-making process. Incorporating eye-tracking into an established behavioral finance experimental framework, we found that making an eye transition between pieces of information that previous research associated with bias makes one less likely to make good financial decisions in future trials. Thus, even the simplest, easy to obtain eye metrics could allow us to more accurately judge if a person’s performance is a reflection of skill, or 16 down to luck and unlikely to be reproduced in the future.