The sample is drawn from a sampling frame constructed from the Sirus inventory using a simple random sampling design stratified by activity, workforce segment and region where possible. The calculation of the allocations combines a Neyman allocation on the rate of innovative units and a proportional allocation on the number of innovative units, which makes it possible on the one hand to be precise on the central variable of the survey (the innovation rate) and also to query a reasonable number of innovative units in order to ensure satisfactory accuracy for questions concerning only these units.