When planning for retirement, deciding what to do with one’s pension funds is a very high-stakes, one-shot decision. In Chile, the process of applying for a state pension requires extensive parsing of websites written in a technical language few people understand. It is therefore not surprising that a majority of citizens in the private pension scheme pay for advice, while many entitled to public benefits leave money on the table by not applying. This unwillingness to autonomously engage with the pension information can have a negative impact on retiree welfare. We argue that one way to address this issue is by reducing the cognitive effort needed to learn about pensions, and we test this idea by experimentally measuring the impact of changing the structure and format of pension information websites on the willingness to engage with the information as presented in tutorials. The baseline treatment websites have a Frequently Asked Questions design (product-oriented) structure with information presented as text, while the experimental treatment conditions have structures including videos and/or oriented to different user profiles. The robustness of the effect of the treatments is compared for individuals learning about public and private plans separately. Process tracing protocols and face-reading analysis to capture the emotional states of the participants when visiting the websites will be implemented. The study is conducted in association with the Chilean Superintendencia de Pensiones and experts from the Instituto de Previsión Social.