Nine ways a question fails before you ask it.

A recon question isn't a survey item and not an icebreaker. It's one sentence whose answer changes what you do with a company, and a call carries two or three of them in a fixed order: one a customer would open with, one that decides the row, one your rep acts on. Almost every bad one fails in one of nine ways, and all nine are catchable before you spend a dial.

The failure modes below come from questions that shipped and didn't work. They are ordered by how expensive the mistake is, which isn't the same as how obvious it is.

The question has no decision behind it

The most common failure and the most expensive, because the call still costs the same. If a yes and a no lead you to do the identical thing, the answer is interesting rather than useful, and interesting isn't a business outcome.

Test it before writing the sentence: what will I do differently for the yes pile and the no pile? If the honest answer is "nothing, but it would be good to know," you've found trivia. Cut it and ask the next thing.

A question earns its dial by sorting. If both answers land in the same pile, you paid to learn nothing.

The list is wrong, so the answer sorts nobody

Every question rests on a structural fact that makes the population candidates in the first place. Ask tire shops whether they've got snacks for customers and the answer sorts served from unserved, because tire shops have waiting areas with people stuck in them. Ask parts counters the same question and almost every answer is no, and none of them are prospects, because nobody sits at a parts counter for an hour with nothing to do.

Same words, same responses, zero value. This one is worth its own page, because it's invisible in the wording: the question reads fine and the whole list is wrong.

Whoever picks up the phone can't answer it

Purchase intent, budget cycles, capital plans and vendor relationships are known by two or three people at a company, and none of them answer the main line. A front desk can't tell you whether the business is considering a major purchase, and no amount of rewording changes that.

The fix is almost never a better sentence. Either ask something the person answering actually knows, or accept that you need a different list reached a different way, which is more expensive and slower, and is a real finding rather than a failure.

It needs thought, or a lookup

The budget is one recall, and it's shorter than people expect. Anything requiring the person to check a system, ask a colleague, or genuinely consider the question has already lost. They will hedge, guess, or end the call, and a hedged answer is worse than none because it enters your data looking like a real one.

"Did your liability coverage already renew this year?" is a recall. "Are you happy with your current provider?" is an opinion that has to be formed on the spot, which is why it produces mush.

The fact isn't the kind a phone can carry

Some things can't be established by voice at all. Whether the shop is well run, whether the owner is open to new ideas, whether the team struggles with scheduling: these are judgments, and the person answering has to form one on the spot about their own workplace. What comes back is a mood. Entered as a yes or a no, a mood looks exactly like a fact.

Ask about a state the business is in. A vending machine is there or it isn't. Coverage has renewed this year or it hasn't. Jobs are scheduled on paper or in a system. If the fact only exists as somebody's opinion of themselves, no wording recovers it, and that's a targeting finding rather than a copy problem: what you wanted to know isn't the kind of thing a call can know.

It telegraphs who benefits from the answer

The moment a question reveals there's a vendor behind it, the answer gets performed instead of reported. People aren't being dishonest; they're managing a sales interaction, which is a different activity from answering a question.

Compare two versions of the same intent:

The left column announces a pitch is coming and gets a defended answer. The right column asks about an observable fact and gets a real one.

The right column is also written as the customer's need rather than the vendor's product, and that is a separate test. "Do you have a vending machine?" doesn't telegraph a pitch, but it is still what a vending company asks; a hungry person asks whether there's anything to eat. Asking for the need is also what makes the answer clean: free snacks on a counter and a machine in the corner are the same answer to the vendor, and a question about the machine would have had to chase which kind. If the line still has your category word in it, it's probably your question in a customer's mouth.

It asks about a need they obviously have

"Do you want more Instagram followers?" "Would more revenue help?" "Are you looking to grow?" Everyone on the list says yes, so the answer sorts nobody. It also does damage on the way out: a question with a foregone answer announces that a pitch is coming, because nobody calls a business to confirm it would like more money.

The fix isn't a reword. It's a different fact. Ask the thing you don't already know, the one that changes how you'd engage them rather than whether they want the outcome. Selling audience growth to realtors, "do you want more followers" is worthless. "Do you mostly represent buyers or sellers" rewrites the whole pitch. Selling to an ecommerce brand, not "do you want to grow on Instagram" but "is Instagram actually producing orders," which is a real yes or no, and the gap it exposes is the reason to call at all.

It's two questions wearing one question's clothes

"Do you work with investor buyers, and do you handle off-market deals?" returns a single yes or no covering two different facts, and you can't recover which. Each half is its own question with its own column, asked in order, and the first answer decides how the second is worded, never whether a new one gets added.

Nobody in particular is asking

The quietest failure, and the one most drafts have. The sentence reads fine, survives every check above, and still gets a receptionist's-voice answer, because nobody in particular is asking it. "You got snacks for customers?" read off a script is a survey. The same words from a 35-year-old with a night side job, a slow leak, nothing eaten since noon and an hour in your bay ahead of her is a question, and a desk answers it the way desks answer people.

Write the question from a specific person with circumstances. Then ask the only test that matters: would the person who picks up tell this caller? If the only person who would plausibly want this answer is someone selling to the business, it's still a sales call, however it's worded, and no rewrite fixes that. The fix is a different asker, not a different sentence.

The circumstances are never spoken on the call. Nobody says "this is Susan from anywhere," and nobody mentions the slow leak. They shape what gets asked and how it sounds; that's the whole of what they do.

A question with no one asking it is a vendor question with the serial number filed off.

The check that costs nothing

Run every draft against the nine before it runs against a list:

1. What do I do differently for a yes than a no? (If nothing, stop.)
2. What structural fact makes these companies candidates at all?
3. Can the person who answers the phone actually know this?
4. Can they answer it from memory, without going to look?
5. Is it a state the business is in, or an opinion they'd have to form?
6. Does the question reveal who benefits from the answer?
7. Would every business on the list answer it the same way?
8. Is it one question, or two?
9. Who, specifically, is asking, and would the person who picks up tell them?

A question that survives review and then wastes three hundred dials costs far more than one rejected in five minutes. Most of the ones that fail in the field would have failed this list on a whiteboard.