"Have we done this before?" is the most expensive question in your company

Every organization has a version of this question. A consultancy asks it when an RFP lands: have we worked in this sector, with this kind of client, on this kind of problem? A clinical research network asks it when a sponsor sends a protocol: have we run this indication, do our investigators know this population, did enrollment hold up last time? A law firm asks it about a clause. An engineering firm asks it about a material.

The question is cheap to ask and expensive to answer, and the cost is almost never counted.

Where the cost hides

The evidence usually exists. It is in a close-out report from 2023, a proposal that lost, a spreadsheet someone maintains, a slide deck from a quarterly review. The problem is not that the organization does not know. It is that the knowledge is spread across formats, folders, and people, and the person who needs it has three weeks, not three months.

So one of three things happens:

• Someone remembers. The answer comes from the person who was there. This works until that person is on holiday, has left, or simply misremembers a number. It is also why "who would know?" is the real first question, and why the same four people are interrupted all day.
• Someone searches. Keyword search across a shared drive finds the files that use the same words. It misses the ones that call the same thing something else, and it returns documents, not answers. Someone still has to read them.
• Nobody answers in time. The proposal goes out generic. The feasibility response commits to a number nobody checked. The clause gets negotiated from scratch.

The third outcome is the expensive one, and it is invisible because nothing visibly fails. The work still ships. It is just weaker than it could have been, and the difference between a qualification summary that cites three comparable engagements and one that says "extensive experience" is often the difference between winning and not.

Why this is hard to fix with search alone

Search finds documents. The question needs an answer assembled from several of them: which project, who led it, what the outcome was, what we charged, what went wrong. That means knowing that "Northgate" in one file and "Northgate Ltd" in another are the same client, that the project in the proposal is the one in the close-out report, that the person listed as lead is the same person in the staff roster.

Those connections are the tribal knowledge. They live in people's heads because no system ever asked for them.

What a real answer looks like

A useful answer to "have we done this before?" has four properties:

1. It names the specific instances, not a count. "Twice: the 2023 pricing project and the 2025 phase-two proposal."
2. It carries the numbers that matter. Budget, duration, outcome, the commitment made and whether it was met.
3. It says why things happened, in the words the files use. "Not selected; budget deferred to next fiscal year."
4. It points back to the source, so anyone can check it in thirty seconds.

That last one is not a nicety. An answer without a source is just another person's memory, and the whole point was to stop depending on that.

What changes when the question is cheap

When the answer takes a minute instead of a week, people ask it more, and they ask it earlier. Proposals get sharper because the comparable work is in front of the writer from the start. Feasibility commitments get more conservative and more credible because the real enrollment history is one question away. Lessons recorded in a close-out report actually get applied to the next project, because the next project's team can find them.

None of that requires anyone to reorganize a shared drive or fill in a form. It requires the files to be read once, properly, and the connections between them to be kept.

That is what Desk is for.