The Epstein paper, that you alluded to, on evaluability assessment should be de rigueur for both researchers and grant officers. In fact I would summarize it as an FPPO cycle. Carry out detailed Formative research or due diligence to prospectively assess the need of the intervention including the ethical issues to rollout, (small scale) Piloting the intervention, not necessarily with a comparator, followed up with a detailed Process evaluation to get a handle on the policies, procedures and key personnel implementing the program's impact in the treatment areas only, and finally a rigorous Outcome evaluation through experimental/quasi-experimental/case control to get net effects etc..
Indeed once an Outcome evaluation is green lighted protocol development including pre-analysis plans and examination of safeguards for human subject safety needs as well as communication/dissemination of study results, data distribution & archiving etc need to be formalized. Such standardized guidelines have been on-going and providing a decent benchmark to assess the readiness of the program staff to undertake an RCT. There was the ASSERT statement checklist which is now subsumed into the SPIRIT statement which may be accessed from:
Note the need for mandatory trial registration. If and when an IRB type body is fully fleshed out within the World Bank, then it would be important for them and concerned public to be able to access a ready cross-referenced repository of all planned, on-going completed and abandoned studies, especially RCTs for periodic audit checks. In point of fact protocols of studies funded through a rigorous peer review process such as NIH/NSF grants should be published in top economics journals, and available open-access online as is done in biomedical journals such as Trials. Accord these the same status and invite the same scrutiny as peer review articles.