In this article, we propose an application of extended Fay-Herriot models with the Generalized Variance Function and the use of benchmarking for estimating short-term business characteristics at a low level of aggregation. This method of estimation, based on the area-level approach, does not require access to unit-level data, which is a clear advantage from the perspective of data protection requirements. The purpose of the study was to estimate the average monthly wage in small companies (employing between 10 and 49 people) in Poland in 2011 for domains defined by the interaction of subregion and type of economic activities. The properties of small area estimators are examined using real data from a short-term business sample survey and administrative registers. Results indicate that the proposed models increase the precision of estimates, which can be used for detailed regional economic studies.