Social learning is important for the life history of many animals, and can lead to stable group traditions and the evolution of culture. To what extent socially learnt behaviours are transmitted, spread and persist will be partly determined by the adaptive use of social learning strategies. However while various learning strategies have been identified in animals, how learning biases interact and co-occur within individuals and populations has rarely been considered¹. In two sets of experiments, we examined the influence of individual-level learning strategies on the spread and persistence of a novel innovation. First, we seeded alternative foraging techniques into eight sub-populations of great tits (Parus major), using a two-action/control experimental design. Automated tracking of individuals revealed that the information spread through treatment sub-populations to reach an average of 75% of individuals (n=414), with a strong bias in each sub-population towards the originally introduced technique². A positive frequency-dependent learning bias (conformist transmission) was observed when first learning, and also in subsequent decision-making. This resulted in an increasing population-level bias to one technique. Second, we varied rewards such that the common technique in each sub-population yielded a lower pay-off than the uncommon technique. Individuals socially learnt to switch to the other technique, providing evidence for a ‘pay-off-biased’ social learning rule. However individuals varied in flexibility, with juveniles and shy behavioural types³ most likely to move from conformist to pay-off-biased copying. Our study thus provides experimental evidence for the integration of two social learning strategies in wild birds, examines individual variation in social learning, and explores the effect of such social learning biases on population-level outcomes.