The call comes at 2:40 on a Tuesday. You are at your desk with headphones on. Your contractor is standing in a half-open kitchen and the sentence begins with "so, we opened up the wall and". There are people on site being paid to wait for your answer. You have ninety seconds of context, no photograph, and a distinct feeling that the more expensive option is the one being described.
You say yes.
Three weeks later, nobody agrees on what you said yes to — the substitution itself, or getting a price for the substitution. Neither of you is lying. Neither of you wrote it down.
That is a communication failure, not a building one — and the part you actually control.
The phone call collapses three separate things
A mid-project decision has three stages: learning what is true, understanding the options, and committing to one. A call from the site fuses all three into a single minute, under time pressure, on someone else's schedule.
Your contractor is not being manipulative by calling. Their crew's clock is running and the fastest way to unblock it is your voice. But the costs of the format land entirely on you: you decide on a compressed verbal summary of something you cannot see, with no time to think and no record afterwards. These disputes are unwinnable later not through bad faith, but because nobody wrote it down while writing it down was cheap.
Ask for the options in writing, then name your deadline
The move is not "I need to think about it." That sounds like a stall and invites pressure. The move is to ask for the same summary you just heard, typed, with three things attached to each option:
- what it costs
- what it does to the schedule
- what it rules out later, once this stage is closed up
Two or three lines per option is enough; this is not a formal document.
Then name the time you will answer by. "Send me those and I'll come back by seven tonight" converts an ambush into an appointment. Most crews take that trade: a predictable delay is easier to plan around than a homeowner who has gone quiet.
You cannot see behind the wall; say so in writing
They are standing in the room. You are not. That asymmetry is the whole situation, and it is worth sorting explicitly:
- Observed — what they can see now that the wall is open.
- Judged — their estimate of what it means for cost and time. Expertise, but still a prediction.
- Assumed by you — that the replacement matches the rest, that the swap is like-for-like, that nothing you already paid for is affected.
Write the assumptions down and send them back for correction. Most bad renovation outcomes are an unstated assumption that survived to the invoice; the cheapest moment to kill one is before the decision.
This is also the line that keeps an AI assistant in its lane. It can help you sort the call, draft the questions, and notice what you failed to ask. It has no idea what is behind your wall. Whatever it produces is a starting point for the person on site to check — never an answer competing with theirs.
Ask what they would do, and what would change their mind
Contractors often present options flatly, because recommending has burned them before. Ask directly: if this were your house, which one, and why?
Then ask the better follow-up: what would make you change that answer?
The reasoning is worth more than the recommendation. A recommendation solves today; the reasoning tells you what this person optimises for — durability, speed, avoiding a callback — and that transfers to every decision left in the project. How to ask for a decision covers the shape of a question that gets an answer instead of a shrug.
Write the decision and the reason where you both can see it
After you decide, send four things: what you chose, the cost and time effect you were told, what it rules out, and the date. Ask them to correct anything wrong.
Most of the value is in the correction. If your understanding is off, you find out that evening, while it is still a text message, instead of in month three when it is an argument about money.
Where a briefing fits
If you are already working the decision through with an assistant, a briefing is that written option set with the ask made explicit and every claim labelled by who made it — you, or your AI.
It does two things here. It gives an asynchronous decision a shape: a stated ask, an honest urgency, the options, and your note in your own words. And the reply lands on the same thread, so what was asked and what was answered sit in one place rather than in your call log. Your contractor needs no account, no app and no assistant — it arrives as ordinary email with a link to the full version and a reply box. Worked examples show the decision-shaped ones.
Honest limits: no delivery guarantee beyond ordinary email. You can see whether the briefing has been seen; you cannot see whether anyone acted on it.
When to just take the call
Some decisions cannot wait an hour. If the crew genuinely cannot proceed, the phone is the correct tool and an hour of paperwork costs more than it saves.
Some contractors will also never answer a written ask. If yours works by phone and always will, adapt rather than fight it: take the call, then send the four-line record afterwards and ask them to correct it. That half you can do alone, and it is the half that survives.
If you would rather your assistant carried the asks that can wait an hour, connect it — the person on the other end needs nothing but an email address.