Your First Build Plan

The best answer: The one piece you could take to a real job tomorrow, on your own phone, and no more.

The piece you can take to a job tomorrow teaches you more in one afternoon than a plan teaches you in a week, because using a thing is how you find out what you wanted from it. Starting with the basement, or with the billing, is a fair choice. It also puts you in front of the hardest work on day 1, with nothing finished in your hands to tell you whether it is any good. A written plan, or a look at what already sells, costs you the week before you learn anything. Asking for all 3 at once leaves you with nothing you can point at.

The best answer: What you would do with it on an ordinary Tuesday, and what you expect to see.

"Keep track of my work" fits a calendar, a spreadsheet, a stopwatch and an invoice book equally well, so any of them is a fair answer to it. Naming your trade, or naming an app you already like, narrows the guess and still leaves somebody else guessing. One ordinary Tuesday, described in what you would do and what you would see, closes it, and nothing in that sentence is technical. Reaching for technical words leads nowhere, because a request that says nothing about your day says nothing about your day in any vocabulary.

The best answer: When I type in 6 today and reload, today reads 6 and the week goes up by 6.

A request you can check has a result you can point at. Quick to open and easy enough for your daughter comes close, because it names a person and something you could watch happen, and it still leaves you nothing fixed to look at tomorrow morning. Safe storage and a named comparison app are real wants that you have no way to call finished. Wanting it to look smart leaves you accepting whatever arrives, and asking for the best technology hands over the one decision that was yours while giving you nothing to check at all.

The best answer: Find out whether the strictest of the 3 can be read at all, before you build.

The unknown that can kill the project is worth answering while the project is still cheap to walk away from. Testing the site she uses least, or asking for all 3 connections at once, does go and find out, and neither one tells you whether the site that decides the project will open its door. A tidy list of every number she wants, or a look at what other sellers pay for, is work you will need later and none of it moves the question. Building the screen first leaves you with a finished screen and nothing to show in it.

The best answer: Deciding what his customers need to see before they will pick up the phone.

What his customers need to see is the decision the rest of the site hangs from, and nothing you hand over can know his trade and his town for you. Choosing which jobs get photographed sits very close to it, and it follows from that first decision rather than the other way round. The contact form and how much of the page fits on a phone each hold a sliver of judgment inside a job that has a right answer. Cropping and typing have a right answer the whole way through. What you keep is the judgment, not the typing.

The best answer: Which numbers your heart doctor asks for, and how often you record each one.

The named numbers and how often you record them are the whole design of the thing. Privacy is the closest of the rest, because it is a real limit, and it shapes one corner rather than the shape of the whole. What the doctors said on the day you left holds those numbers somewhere inside it, in a form nobody can build from. Saying that you are new, that your week is busy, or that it should take care because the subject is health, all describe you rather than the work. None of them change a decision it has to make.

The best answer: Start a new conversation for the sign-in page and leave the invoice hour behind.

An hour of invoice detail sits between you and the new question, and none of it helps with a sign-in page. Having it write down what you settled is worth doing on its own account, and the conversation is still carrying that hour afterwards. Telling it to forget does not empty what it is already holding, and finishing every last invoice detail first is how 1 hour becomes 3. Everything it holds onto competes for its attention, so carrying on in the crowded conversation, or asking for both screens inside it, is where the drift costs you most.

The best answer: Renaming 600 receipt files that follow no pattern, which you do every March.

The receipts repeat, the rules can be written down, and a glance tells you whether the result is right. An apology draft is the next best, because you decide what goes out and nothing leaves your hands until you send it. Comparing 3 quotes, or working out next year's prices, has real arithmetic in it. The decision underneath turns on things the numbers do not carry, so a handed-over answer arrives sounding more settled than it is. Letting clients go and cutting somebody's hours cost you something you cannot take back.

The best answer: Book yourself in, the way a customer would, and see what the screen does.

A report that the work is finished is a claim, and people who use these tools every day describe unfinished pieces sitting inside work that was announced as done. Tests are the strongest of the rest, and they are written by the same hand that decided the work was finished, so they carry what it already believed. Asking whether it is sure, reading every line, or glancing at the page each cost you time, and none of them puts the form through what a customer would do to it. Moving straight on is how the gap reaches a customer before it reaches you.

The best answer: A change in one place can quietly break another, and your own check catches it.

The calendar may well be perfect. Asking for the confirmation email to be left alone is a real technique. It works right up until you meet the part you did not think to name, and that is a list you can never finish writing. Suspecting the email was broken already, or blaming your own wording, sends you looking where the evidence does not point. The habit worth building is checking the things around your change rather than only the change itself, and neither asking for the whole thing again nor starting the page over builds it.

The best answer: Stop, go back to the last version that worked, and describe the problem again.

Each of the 6 failed tries has left something behind, so by now you are untangling 5 earlier attempts as well as the original problem. Sending the exact error is the best thing you can do inside a working debugging loop, and here it aims a good technique at ground that stopped being solid several tries ago. A new approach every try, or building round the feature altogether, both accept the mess rather than clearing it. Going back to the last version that worked costs you 6 tries instead of the whole project, and people report 1-line fixes hiding behind exactly this pattern.

The best answer: What does the app need that sits on your own laptop and nowhere else?

Something works in one place and not another, so the difference between the 2 places is where the answer lives. The account she was never given and the untested browser are both real differences and both worth asking about, and each one is a single guess at a question you could ask in general. Wondering whether the code is wrong points at the one part that is identical in both places. Slow internet and trying again in an hour leave you waiting for the problem to explain itself.

The best answer: Saving a copy of the working version you can go back to, before you start.

A saved copy is the only one of these that survives a change going wrong. Working one small piece at a time is the strongest of the rest, and it shortens the distance back to a working app without giving you any way to travel it. A written description of the change, or a list of what the change will touch, are claims about the future rather than a copy of the present. People describe losing a full day of work to a change they could not undo, and asking it to be careful is the wording that protects nothing at all.

The best answer: Ask what is inside it and why it is not needed, then make the call yourself.

You are being asked because this one cannot be taken back, and one question costs you 10 seconds and tells you what you are deciding about. Moving the folder yourself comes closest, because it turns something you cannot undo into something you can, and it spends an afternoon of waiting on what a sentence would have told you. A saved copy is a reason to be calm rather than a reason to skip the question, and refusing everything forever costs you the reason you started. Allowing it and looking afterwards is looking after the folder is gone.

The best answer: There is a part of your own product you cannot fix or defend when it breaks.

Understanding is what you fall back on at 2 in the morning, and it matters here for that rather than for pride. Notes on that part are worth having, and the gap they close sits in the file rather than in your head. Wishing you had typed it yourself, or stopping all new work until you have read everything, both take the problem seriously and neither one is a way to carry on. The day it breaks is the worst day to start reading, so what you want is to be able to say what that part does and roughly how, before you need it.