One Question, Forty-Eight Hours

Most genuinely important questions are narrow. A point of regulatory mechanism. A single jurisdictional fact. Whether a stated position means what it appears to mean. One question, needing a proper answer, once.

The advisory industry cannot sell that, and the reason is structural rather than greedy.

Why the industry cannot price a single question

Firms are built around engagements because engagements are how the economics work. An engagement amortises the cost of understanding a client's situation across a body of work. It supports a team. It justifies the partner time spent winning it. It produces a deliverable substantial enough to be circulated internally by the person who commissioned it, which is frequently what they actually need.

A single question defeats every one of those. The context cost is the same and cannot be spread. There is no team to occupy. The winning effort exceeds the fee. And the output is a paragraph, which is difficult to circulate as evidence that money was well spent.

So a principal holding one narrow question faces two options: commission something several times larger than required, or ask nobody. In practice most ask nobody — or, increasingly, ask a machine, which is the same as asking nobody but faster and more confident.

What a narrow question actually requires

The reflex is to treat this as a retrieval problem, and retrieval is now close to free. That reframing is wrong about the work.

A narrow question is rarely a request for a fact that exists somewhere and needs finding. It is usually one of three things. It is a question about mechanism — not what a rule says but how it operates, which is a matter of practice and sequence rather than text. It is a question about significance — whether a development means what it appears to, which requires knowing what the alternatives would have looked like. Or it is a question about whether the question is right, which is the most valuable category and the one no retrieval system can serve, because answering it requires telling the enquirer that they have asked about the wrong thing.

None of those is solved by having more material available. They are solved by someone who has done the work before recognising the shape of the situation — and the scarce input is judgement about which of the available facts bear on it.

This is why the human commitment matters and why it should be defended without disparaging the technology. Retrieval is genuinely cheap now, and it is genuinely good at retrieval. The answer to a narrow question is almost never a retrieval problem, and a confident synthesis of everything published about a topic is exactly the wrong output when the useful answer is that the published material addresses a different question.

The most valuable answer is often a different question

The third category above deserves separating out, because it is where the largest share of the value sits and it is the one a client never asks for.

Questions arrive shaped by what the enquirer already believes. A principal asks whether a particular filing obligation applies to a particular vehicle, because they have identified that as the exposure. The useful answer is sometimes yes or no. It is at least as often that the obligation is minor and something adjacent is not — a different deadline, a different entity in the structure, a consequence in another jurisdiction that the framing excluded.

Answering only what was asked is the safest thing a provider can do. It is precisely responsive, it is defensible, and it leaves the enquirer with a correct answer to a question that was not the one that mattered. Every incentive points that way: the wider answer takes longer, was not commissioned, and carries the implication that the enquirer misdiagnosed their own situation.

Which is why this has to be a deliberate part of the product rather than a courtesy. A response that says the answer to your question is X, and the reason you are asking it suggests you should also know Y is doing something a retrieval system structurally cannot, because the system has no view about why the question was asked. Whether the reformulation is welcome is a separate matter, and it is not always.

The mechanics, and what they concede

The service is deliberately shaped to fit the case: a question is submitted, reviewed, and a price sent back; on approval, a focused response written by a human arrives within forty-eight hours.

Two features of that are concessions rather than selling points, and they are the honest part.

The price follows the review, not the submission. Nobody knows what a question costs until someone has read it, because the same number of words can describe an afternoon's work or an unanswerable problem. A firm quoting before reading is quoting for a length of document rather than for an answer.

Some questions are declined. A question may be unanswerable at proportionate cost, or answerable only by someone in a jurisdiction where we hold no tested capability, or actually a request for legal or investment advice wearing a different hat. Declining is the correct outcome in each case, and it happens often enough to be worth stating in advance rather than discovering.

The forty-eight hours is a constraint as much as a promise. It bounds what can be attempted, and a question that genuinely needs three weeks is not improved by being answered in two days — which is another reason the review stage exists.

The wider observation is about what gets asked at all. An industry that can only sell large engagements has quietly trained principals not to ask small questions, and the questions that go unasked are not the unimportant ones. They are the ones that seemed too minor to justify commissioning anybody — which is a category that includes a substantial share of the decisions people later wish they had checked.