Most commissioned research is used once because what gets stored is the conclusion rather than the evidence. A finding without its screen, its question set and its date cannot be reused — the next team cannot tell who was asked, what they were asked, or whether it is still true.
Why is most research used once?
Because firms store outputs and discard inputs. The deck survives; the screen that defined who counted as an expert, the questions asked and the raw record do not.
Without those, a later reader has a conclusion they cannot evaluate, and the rational response to a conclusion you cannot evaluate is to commission the work again.
The economics of that are worse than they look, because the second commission is usually not cheaper than the first. Nothing about having done the work before makes doing it again faster if the screen and the questions were not kept, so a firm can pay full price for the same finding several times and record it each time as a new project.
- 01
Capture the screen
Store who counted as an expert and why, not only who was interviewed.
- 02
Record the questions
Keep what was asked, not only what was answered.
- 03
Tag by question
File against the question the research answers, not the project that commissioned it.
- 04
Date everything
Attach a date and a decay expectation to every finding.
What has to be captured?
Four things: the screen, the questions, the raw record, and the date. Those make a finding checkable by someone who was not there, which is the definition of reusable.
None is expensive to capture at the time and all are effectively impossible to reconstruct afterward.
The date is the item most often skipped and the one that determines everything else. A finding with no date cannot be assessed for decay, which means a later reader has to treat it as either current or worthless, and in practice treats it as worthless. Stamping the fieldwork window costs nothing and is what makes the other three usable.
When does capture have to happen?
At the time. Reconstructing a screen six months later is unreliable and usually not attempted, which is why archives fill with conclusions nobody trusts enough to reuse.
Making capture a deliverable rather than an aspiration is the only reliable mechanism.
Making it a deliverable also has to mean it is checked. A capture requirement nobody reviews degrades to a folder of transcripts with no screen attached, which is the most common half-implemented version and produces most of the cost with little of the benefit. The check can be brief; it cannot be absent.
| Project archive | Reusable research asset | |
|---|---|---|
| Stores | Outputs, filed by project | Screen, questions, record and date, filed by question |
| Answers | What did that project conclude | What do we know about this market |
| Reused | Rarely, because it cannot be evaluated | Routinely, because it can be checked |
How does the next team find it?
By question, not by project. The next team is looking for what is known about a market, not for what a named project concluded, and project-based filing makes that search fail.
That is a taxonomy decision made once and lived with for years by every team that inherits it, which is why it is worth making deliberately.
Retrieval also has to survive people leaving, which is the real test of a taxonomy. A structure that works because the person who built it remembers where things are is not a structure, and it fails silently at exactly the moment it is most needed — when the person who ran the original project is no longer there to ask.
How fast does research decay?
Unevenly. Structural findings about how a market works hold for years; anything about pricing, capacity or competitive position is usually stale within a few quarters.
Tagging findings by type at capture is what lets a later reader tell which is which without re-reading everything — the same recency discipline a bench needs.