We study a theoretical model of behaviour that introduces as a primitive a dataset of “approvals” for objects appearing as a list. Approval situations are typical of online behaviour. Approval is distinct from choice as it does not guarantee a final choice (e.g. when filling a virtual shopping cart or selecting potential partners on a dating site) or it may not involve a final selection at all (as e.g. in social media interactions, when retweeting or reacting to a Facebook post). We study the identification, characterisation and comparative statics of the model, and we introduce the problem of “list design”, whereby a designer of lists can manipulate an agent’s choice to maximise some objective function (e.g. number of clicks).