I completely support the idea that a failure, whether from a roll, player decisions, or plain old GM fiat (there isn't enough to make that clue out YET) shouldn't necessarily be the end. I have had good success with taking a failure and using it to add a dimension to later success, often including that aha moment of what it was that caused the failure before, whether it was chance, poor skill, or lack of information to make the important details stand out.
In a way, as long as you encountered Detail A, whether you succeed in revealing subsequents A1, A2 and A3 right then, or later doesn't always matter. But the act of looking at Detail A does, and hopefully that give the characters, and their players, a direction to think in.
In a "logical sense", it sounds like your approach is let them roll until they succeed, and once that isn't viable, have a fallback plan that goes into definite failure but still leads somewhere. I /think/ that this is much more compatible with the GumShoe idea if it was packaged right, than your think.
I really think the GUMSHOE approach (for those who don't know,. you have 0-10 points in 30+ forensic skills, and if you want to be the one to find a clue in a situation, you spend a point and find it. What it means may or may not be included, but what it means in context of the whole mystery is usually left to the players to work out) is about spotlight sharing. It's my turn to show I can find a difficult fingerprint anomaly. Otherwise the team could just have an even spread across all things, and the GM could turn it into an OOC guessing game by having more clues than they have points in a give forensic skill, and if the players don't choose to spend when important, too bad.
So, I think you could really read it as "Chararacters never miss a chance to interpret a clue, though they may fail if they are missing another clue to add context to the first." After that, its about players talking it out, and giving skill rolls to understand what a clue means, or where it leads.
Specific example: Locked doors and windows, no finger/glove prints on anything. I know that a wall crawling spider themed villain was in the room, and stuck to the ceiling, lowered themselves down on a web, and did X, then went up and crawled through the ceiling panels the same way they came in. IF a player asked, they could quickly find clues on the ceiling. If they had no reason to look, their rolls would just confirm the details. No nothing was cleaned off, there are no impressions on the carpet even. A remote drone would have disturbed the papers on the desks, etc. Later on, finding some web, or whatever, they might think to check the ceiling. I might even allow that with new info their skill could suggest check the ceiling.
Framing is important, especially when the characters are supposed to be talented pros like on Star Trek. (This is where limiting margin of success by how many points were in a skill made characters who invested in a skill feel like pros).