It’s also probably worth pointing out that medical interventions typically go through multiple phases of clinical research.
Preclinical research is conducted to see if there’s good reason to think there’s anything to research.
Phase 0 trials are conducted on a handful of volunteers, basically just to see how to administer the treatment (to make sure an oral drug is actually digested and becomes bioavailable rather than simply passing through, for example).
Phase I trials are conducted on a larger number of people, mainly to try to work out dosages.
Phase II trials are where you start to see RCT (Randomized placebo-Controlled Trials), but not necessarily. They’re for initial assessments for safety and efficacy.
Phase III trials are the big time. This is where you recruit large numbers of subjects, and conduct large scale RCTs to figure out how well the treatment actually works.
Phase IV trials are basically just ongoing monitoring of actual usage “in the wild.” In a sense, all medical interventions are experimental - individuals and disease progression between individuals vary far too much to ever be able to say with 100% confidence how a given treatment will work.
So, by Phase II, and certainly by Phase III, when you actually use placebos, the researchers should have some idea of how effective a treatment will be, and how dangerous.
As Chronos rightly points out, a treatment may not just wind up being useless, it might be worse. In the infamous recent example, hydroxychloroquine for treating COVID-19, some preliminary studies showed some effects, but the Phase II and Phase III RCT trials pretty much all showed either no clinical effects, or worse clinical outcomes for those receiving HCQ.