Saying no to a hopeful client

Their optimism is usually built on something true — keep the true thing, and only take away the conclusion.

The idea

When a client is more hopeful than the facts support, they have almost always noticed something real. If your correction quietly deletes that real thing, they stop hearing you and start defending. If you agree with everything to stay pleasant, you have simply moved the bad news to a worse date.

The way through is narrow and learnable: agree with the fact, disagree with the inference, show why, say your own view, hand them one thing to do. Order matters more than wording. The same five sentences, rearranged, can read as generous, cold, or spineless.

Build the response

Pick the moves in the order you'd actually say them, then play it back. The line tracks a rough trust index (0–100): how much of your next sentence the client is still willing to take in.

try an order:

How it works

Five moves. Each one does a job the others cannot do, and each one is cheap — a sentence or two. The default order is the one that keeps the client in the room.

1  acknowledge  "The side letter is real, and it matters — the strongest
                document in the file."          (their find survives)
2  reframe      "What it settles is that a promise was made. What it
                doesn't settle is whether the loss flows from it."
3  evidence     "Two of the four heads of loss rest on records we don't
                have — that's where the last three matters turned."
4  recommend    "My working view is nearer sixty than ninety, and I'd
                budget for a contested hearing."
5  next step    "Two ledger exports by Friday; I re-run it with you Monday."

Why this order. The acknowledgement is free before the correction and worthless after it — said late, it sounds like consolation. The reframe is the load-bearing move: it is the only sentence that lets a client hold "my document is strong" and "my conclusion was wrong" at the same time. Evidence before the reframe is a data dump; after it, it explains a gap the client has already accepted. The recommendation is what you are paid for — a range with a reason, not a hedge. The next step is what stops honesty reading as abandonment.

And when they say “the other advisor said 90%”

Do not audit the other advisor. The moment you comment on their competence or information, the client has to choose a person, and people defend the person who agreed with them. Compare questions instead.

don't  "He probably hasn't seen the quantum papers."   (attacks the person)
don't  "Well, everyone has a view."                    (says nothing at all)

do     "Ninety is right for the question he answered — do they owe
        something. Mine is a different question — how much we recover,
        and when. Ask him which one his ninety is; I'd like to compare
        like with like, and I'll send you the two questions today."

Three things happen at once: his number stays intact, yours becomes comparable rather than contradictory, and the client leaves with a task that resolves the difference instead of adjudicating it.

When to use it

Situation	How it lands, and the trade-off
The optimism rests on something genuinely good	Ideal. The acknowledgement is true, so it costs nothing. Trade-off: it takes four minutes rather than one, and some clients hear the first sentence as agreement — the reframe must follow immediately.
A second opinion is more optimistic	Works well if you can name the different question the other view answered. If the gap is genuinely an error or missing information, this frame will not carry it — say so directly to the client and take it up with the other advisor, not through the client.
A decision is imminent, or there is a deadline or safety risk	Compress: one clause of acknowledgement, then the recommendation first. Speed beats sequencing when the next hour matters.
The optimism has no factual basis at all	Do not manufacture an acknowledgement — invented agreement is heard as technique. Acknowledge the instinct instead ("you're right that waiting is expensive") or skip straight to the reframe.
Advice where the range itself is the product	Put the range and its basis in writing the same day. Spoken reframes drift; by the board meeting your "about half" becomes "he said it could be fine".

Watch out for

The acknowledgement that isn't one. "I hear you", "I understand your frustration" and "that's a fair point" are labels, not agreement. Name the specific thing they got right — the document, the clause, the instinct — or don't bother.
The word "but". Everything before "but" is deleted on arrival. Use a full stop and start a new sentence: "The side letter is strong. The open question is causation."
Numbers used as camouflage. "Sixty to seventy percent, roughly, but it's hard to say" is still a no, delivered badly. If you can't defend the range, you're hedging, and clients can hear it. Give the number, then the reason, then the thing that would move it.
Any comment on the other advisor. Even the mild ones — "he may not have the full picture" — are heard as a status move. Compare the two questions, never the two people. If the client presses, offer a joint call: confident people are happy to be compared side by side.
Ending on the bad news. Accurate and actionless reads as abandonment, and it's the most common failure in the room. One task, one date, one owner, said last — that single sentence does more for trust than any amount of extra evidence.

Worked example

Say the interviewer gives you this: your client is convinced a disputed supplier claim is "ninety percent, easily", because a colleague put it that high and because they've just found a signed side letter. You think recovery is closer to half, and slower.

You open with the true thing: the side letter is the strongest document in the file, and on liability alone you'd be near ninety yourself — which is very likely the question their colleague answered. Then the separation, in one sentence: liability and recovery are different questions, and yours is about how much comes back, and when. Only now the specifics — comparable matters recovering forty to seventy percent of the claimed figure, two heads of loss resting on records nobody has yet. Then your own view, unhedged: plan on about half, and give the board that range now rather than in March. Finally, the task: ask the colleague which question his ninety answers, and you'll send the two questions today so the comparison is like with like.

Notice what you never said: that the colleague was wrong, that the client was naive, or that it's "too early to tell". The client leaves still holding their good document, now holding a realistic number, and with something to do before Friday. That is the whole skill.