Household finance surveys are widely used for policy-making. Understanding the wealth distribution is important to have a better understanding of how households respond to shocks, to assess their level of indebtedness and to estimate the effects of taxation and redistribution policies. Most of the Central Banks in the Euro area collect micro data on household finance and consumption (Eurosystem Household Finance and Cnonsumption Survey, HFCS). For instance, Bance d'Italia conducts the Survey of Italian household income and wealth (the Italian HFCS) since 1965. Yet, the measurement of household wealth through sample survey is a difficult task. Wealth is usually very concentrated in the hands of few households who also have in general a lower propoensity to accept the interview. Unfortunately auxiliary information on household wealth is rarely at the design stage. In this paper, we describe the results of a joint project between Istat and Banca d'Italia whose aim is to modify the sampling design of the Italian HFCS by explointing the administrative information coming from personal tax records. We discuss the methodological challenges to be dealt with when using income as a proxy variable for household wealth. We then present the optimal stratification and sample allocation based on the genetic algorithm by Ballin and Barcaroli (2014). Finally we estimate the benefits of the methodlogy in terms of bias-reductiona dn increase of precision of the estimators.