Every one of them had a written position or sequence that could be argued with before it was produced at scale. Every one of them had someone internal named as owner of the result. And in four of the five, the first proposed answer was revised before it was implemented.
That last number is the honest version. We do not claim a hit rate, because the outcome depends on decisions the client makes after we leave. What we can describe is the process that produced a revised answer in four of five cases: an internal reviewer who was not the author, and a scheduled point at which the plan was reopened.
Both of those are cheap to set up and both are hard to fake. If a consultancy cannot describe who reviews its own first draft, or when the plan gets reopened, the correction has probably never happened on that engagement.