Accounts of addiction are highly prevalent in online forums and community message boards. This rich database provides an unprecedented opportunity to find out exactly how people change their addictive behaviour (including rare or under-reported behaviours such as internet pornography and gaming). Guidelines for changing addictive behaviours have started to emerge which are founded on lived experience. The methodology for identifying, extracting and synthesizing this data from online sources is however extremely time and resource intensive. The current study proposes a new approach involving supervised machine learning and natural language processing. To do this, we will develop a programme which can efficiently detect the mechanisms of change reported by millions of consumers. This work has the potential to expand consumer voice in the development of interventions by acting as an intermediary between consumer to consumer wisdom.