Digital contact tracing applications have been introduced in many countries to aid in the containment of COVID-19 outbreaks. Initially, enthusiasm was high regarding their implementation as a non-pharmaceutical intervention (NPI). Yet, no country was able to prevent larger outbreaks without falling back to harsher NPIs, and the total effect of digital contact tracing remains elusive. Based on the results of empirical studies and modeling efforts, we show that digital contact tracing apps might have prevented cases on the order of single-digit percentages up until now, at best. We show that this poor impact can be attributed to a combination of low participation rates, a non-flexible reliance on symptom-based testing, low engagement of participants, and delays between testing and test result upload. We find that contact tracing does not change the epidemic threshold and exclusively prevents more cases during the supercritical phase of an epidemic, making it unfit as a tool to prevent outbreaks. Locally clustered contact structures may increase the intervention’s efficacy, but only if the number of contacts per individual is homogeneously distributed, a condition usually not found in contact networks. Our results suggest that policy makers cannot rely on digital contact tracing to contain outbreaks of COVID-19 or similar diseases.