We examine non-response error in expenditure data collected with a mobile app: 2,432 members of the Understanding Society Innovation Panel were invited to download an App to record all their spending on goods and services for a month, by scanning receipts or reporting spending in the app. We examine participation at different stages of the data collection, including the effects of an incentive experiment. We then examine biases in the types of sample members who participated, considering social-demographic characteristics, financial position and behaviours, mobile device ownership, usage patterns, self-rated skills and indicators of cooperativeness collected in prior panel waves. We also examine the quality of expenditure data collected by comparing estimates of monthly individual and household expenditure with data from previous waves and with self-assessments of expenditure missed. We assess process quality by examining the pattern of receipts returned over time, spending occasions verified by a scan and the quality of coding.