• Photo uploads to social media are linked to specific uses and are not a good indicator of subjective values. • Landscape photo composition is not linked to preferences. • A dedicated mobile app provides useful insights into in situ location specific perception and preferences, especially with regard to quieter values of urban green space. • Combined with observations of use a dedicated app can help prioritise locations for improvement. Subjective values of urban green spaces are difficult to quantify and thus easily overlooked in planning processes. Accounting for such values is an important challenge in developing sustainable cities. Crowdsourcing methods, such as big data and smart phone applications, have emerged as promising methods to improve insights into subjective perceptions and preferences. However, we know little about how well these relatively new methods actually quantify subjective values. We assessed several of these new methods by comparing observations of use (n = 1009) to three crowdsourcing methods in one large park in Amsterdam, the Netherlands: a dedicated mobile app providing in situ stated preferences (n = 377), passive social media (n = 78) and a municipal reporting app (n = 187). We show that observed use and passive social media only captured user quantity and were not able to identify green space qualities that are important for mental health functions, such as how relaxing or safe a location is. The dedicated mobile app combined with observed use helped to identify priority locations for improvement. Our findings emphasize that if inadequate measures are used in smart city developments, subjective values and specific user groups will continue to be overlooked in planning processes.