Household surveys are widely used as a tool for obtaining information on people's socio-economic status and well-being. However, the accuracy of household survey estimates decreases significantly when it comes to making inferences for population groups who represent disaggregations for which the survey was not designed. It is possible, in this context, to use estimation processes that combine information from household surveys with existing auxiliary information at population level, such as censuses or administrative records. This paper offers a methodological guide to the combination of survey statistical techniques with probabilistic models in order to produce disaggregations for interest groups, known as small area estimation (SAE) techniques.