Research Credit Guarantee
Each of these is a state the system records for itself, so the refund is something you can check on the account rather than something you have to argue for.
We took the credit, ran the research pass, and could not produce a usable result. That is our failure, not a finding about the company.
We researched the company and found no usable recipient — no published address, and nothing our contact search could stand behind. An account you cannot email is not one you should pay for.
The research landed but we could not turn it into an email worth sending to anyone on the account. The research is kept; the credit goes back.
Sometimes deep research rules an account out on a fact that was already in front of us before we spent your credit. When the deciding fact was one we already held, the credit was ours to waste, not yours.
That one is not refunded, and we would rather you read it here than discover it later. When research does its job and the honest answer is no, the credit bought exactly what it was for: a decision, with the evidence behind it, before you spent a message and a prospect’s patience finding out the hard way.
The distinction we hold ourselves to is whether the work was wasted or merely disappointing. A well-evidenced no is not a failure to deliver. A credit spent to learn something we already knew is.
Refunds are automatic. The credit returns when the failure happens, so there is never a credit you are owed and have not been given.
A refunded credit lasts thirty days from the day it is returned, which lets it outlive the monthly reset. We spend whichever credit expires first, so neither pool is stranded while a longer-dated credit is used.
An account can be refunded once. Beyond that, one failing the same way repeatedly would hand its credit back forever without ever consuming one.