Researchers, planners and policymakers in low- and middle-income countries often lack up-to-date and representative data about urban populations. Censuses are infrequent and expensive, while periodic surveys generally rely on sampling frames derived from censuses. In contexts of rapid population change, this can limit the quality and value of the data collected. Moreover, even well-executed survey exercises can result in systematic biases or exclusions of certain groups.

Based on more than 50 years of combined experience of conducting survey research in a wide range of contexts, we offer practical reflections and suggestions for overcoming some of these problems. We discuss issues related to research design, sampling in cities without reliable census data or lists, practical issues associated with the design of survey instruments and the process of enumeration.

We conclude with ideas for reducing the time and costs associated with gathering high quality data in future survey exercises.