Although stated-preference surveys take various forms, the use of either text or visuals to represent attributes is uncontroversial, and they remain the commonly used formats. While prior research has investigated the impact of these commonly used formats in other disciplines, little is known about their effects on results in terms of relative importance in environmental economics literature. We conduct surveys on households’ preferences for water efficient technologies in South Africa, where we compare three presentation formats, namely text, visuals, and both text and visuals. Survey data collected from 894 heads of households in the Gauteng province is analysed using the mixed logit model to test whether these three formats generate differences in estimated utilities and willingness to pay. This research sheds light on how to develop a valid presentation method for attribute levels in choice experiments, which is critical considering most environmental economics goods and services are not traded in the market. Our results show that the visuals format generates more statistically significant coefficients than the other formats. This suggests that the presentation format has significant impacts on choice. The choice between the three elicitation formats may imply a trade-off in choice precision. Our findings suggest that more research on presentation formats in environmental economics is warranted.