Inflation expectations are a key variable in conducting monetary policy. However, these expectations are generally unobservable and only certain proxy variables exist, such as surveys on inflation expectations. This paper offers guidance on the appropriate quantification of household inflation expectations in the Swiss Consumer Survey, where answers are qualitative in nature. We apply and evaluate different variants of the probability approach and the regression approach; we demonstrate that models which include answers on perceived inflation and allow for time-varying response thresholds yield the best results; and we show why the originally proposed approach of Fluri and Spörndli (1987) has resulted in heavily biased inflation expectations since the mid-1990s. Furthermore, we discuss some of the key features of Swiss household inflation expectations, i.e. the fact that there has been a shift in expectation formation since 2000 (expectations are better anchored and less adaptive, and there is lower disagreement of expectations). We suggest that this may be linked to the Swiss National Bank's adjustment of its monetary policy framework around this time. In addition, we outline how expectation formation in Switzerland is in line with the sticky information model, where information disseminates slowly from professional forecasters to households.