Skip to content
QDNALearn AI, from beginner to expert
FR

Lesson 7 · Intermediate · 15 min

Iterating: the first answer is a draft

Iterating a prompt in Copilot Chat: the first answer is a draft. Precise follow-up, critique of the answer, before and after example, exercise and quiz.

Goal
You will be able to follow up in Copilot Chat with a precise critique, changing one thing at a time, until the answer is usable, without starting over.
Skills
Steer
Three successive versions of a text in a Copilot Chat window, each linked to the follow-up that produced it, from first draft to final version.
Illustration generated by AI

Your first attempt, unaided

Take an answer from Copilot Chat that did not suit you and follow it up with the first sentence that comes to mind. Compare the new answer with the previous one: what improved, and what disappeared?

In brief.

Copilot Chat keeps the thread of the conversation. The first answer is a first draft: you critique it precisely instead of redoing it. A good follow-up says what to keep, what to change and what the expected result looks like. You change one thing at a time, you check, then you follow up. Three well-run turns beat an imaginary perfect prompt.

  1. 1What a follow-up must say to be obeyed

    Microsoft writes that the first answer is a draft and recommends continuing the conversation: "make it shorter", "use a more formal tone", "add a column on risks". Each turn keeps the context of the previous turns. So you do not have to repeat the whole prompt: you adjust.

    Iterating is steering. The word iteration names this cycle: read the answer, judge it against your goal, formulate a correction, follow up. The skill at play is called steering: redirecting the model without losing what was already good.

    An effective follow-up has three parts. What to keep: "keep the structure and the first paragraph". What to change, in checkable terms: "80 words at most", "start with the key figure". And, if possible, an example of the expected result. "That's not it" teaches the model nothing; "the second bullet is off topic, tighten it around the budget" puts it back on track.

    Weak follow-up Strong follow-up
    "Make it shorter" "80 words at most, keep the three main ideas"
    "Too technical" "Replace every technical term with an everyday expression or an explanation in brackets"
    "That's not it" "The second bullet is off topic; tighten it around the budget and the deadline"

    Two reflexes complete the method. Change only one thing per turn: if you alter the tone, the length and the format at once, you will not know what produced the improvement. And ask Copilot for its assumptions before it produces anything: "list what you are assuming about my request, then answer". You correct the wrong assumptions before it builds on them.

    One reservation, finally. Asking Copilot to "check its answer" without giving it a source rarely improves the facts: research shows that models correct themselves poorly on reasoning on their own. The useful critique comes from you, or from a confrontation with a provided document.

    Diagram "The first answer is a draft: iterate": 1. Complete prompt; 2. Copilot's answer; 3. Check: facts, sources, tone; 4. Follow up: "shorter", "as a table", "cite your sources"; Result validated by you. Note: Trap: copy-paste the first answer without reading itDiagram "The first answer is a draft: iterate": 1. Complete prompt; 2. Copilot's answer; 3. Check: facts, sources, tone; 4. Follow up: "shorter", "as a table", "cite your sources"; Result validated by you. Note: Trap: copy-paste the first answer without reading it
    Diagram "The first answer is a draft: iterate"Diagram generated by AI and reviewed
  2. 2An expenses announcement shortened without loss

    An internal communications officer wants to announce a new expense claim procedure. The first draft is too long and too technical.

    Weak prompt, as a follow-up:

    It's too long and too complicated, redo it.
    

    Summarised answer: Copilot shortens at random, keeps the jargon ("approval workflow") and drops the effective date. It guessed what "complicated" meant.

    Strong prompt, as a follow-up:

    Keep the three-paragraph structure and the 1 November date. Rewrite in 120 words at most for colleagues who have never used the tool. Replace every technical term with an everyday expression or an explanation in brackets. Start with what everyone must do from their very first expense claim.
    

    Summarised answer: three paragraphs, 110 words, the date kept, "approval workflow" turned into "approval chain (your manager, then accounting)", and the concrete action up front.

    What changes: the follow-up names what to keep, so nothing important disappears. The corrections are numbered and checkable: you can count the words. The recipient is restated, which guides the choice of vocabulary.

  3. 3Rewrite an invoice follow-up in three parts

    You asked Copilot Chat for a reminder email to a client who has not paid a 3,400 euro invoice that fell due twelve days ago. The first draft runs to 220 words, addresses the client too casually and mentions penalties that your contract does not provide for. Here is the starting follow-up, to be improved:

    "No, that's no good, make it more professional."

    Rewrite this follow-up in three parts: what to keep (the invoice number, the amount, the request for payment within eight days), what to change (formal address, 90 words at most, no mention of penalties) and an example opening line. Run it, then do a second turn that changes only one thing. The composer below helps you word the initial prompt; the follow-up is written directly in the conversation.

    Self-assessment grid: (a) every follow-up says what to keep and what to change; (b) only one thing changes per turn; (c) the final email fits in 90 words and mentions no penalty.

    Open the prompt composer

  4. 4The long conversation that mixes up instructions

    After some forty exchanges in the same conversation, Copilot mixes up the instructions of two tasks and forgets the format requested at the start. The context piles up and gets diluted: the initial instruction is no longer dominant, and "like earlier" no longer points to anything precise. Correction: one conversation per task. When you change topic, open a new conversation or write "new topic", as Microsoft recommends. For an important follow-up, paste the complete prompt again, block by block, with the latest version of the text, rather than referring back to a distant exchange. Copilot Chat keeps your conversation history: you can find the old one without extending it. Rule to remember: one conversation per task, one complete prompt per important follow-up.

  5. 5Quiz

    Three questions, instant feedback. Each option comes with an explanation.

    1. Which follow-up is most likely to improve an answer that is too long?

    2. After a long conversation, Copilot mixes up the expense claim and the meeting report you were working on before. What do you do?

    3. You follow up with "make it shorter". According to the four-ingredients lesson, which element is missing?

  6. 6Proof of mastery

    Paste your starting prompt, your two follow-ups and the final answer obtained in Copilot Chat.

    Intermediate badgeThis lesson counts towards the Intermediate badgeSee the four badges

    Criteria

Going further

The lesson on the four ingredients recalls what a first prompt must contain; a follow-up often corrects a forgotten ingredient. The term iteration is defined in the glossary. The next lesson, attaching a Word, PDF or Excel file, shows how to give Copilot a source to confront its answers with.

Frequently asked questions

Do I have to repeat the whole prompt at every follow-up?

No. Copilot Chat keeps the context of the conversation, and Microsoft recommends continuing the exchange with short requests. Repeat the full prompt only when the conversation has grown long or when you switch tasks.

How many turns before giving up?

Three precise follow-ups are usually enough. If the third improves nothing, the problem lies in the initial prompt: an ingredient is missing, often the source or the recipient. Open a new conversation with a complete prompt.

Is asking Copilot to check its own answer enough?

Not for facts. Research shows that models correct themselves poorly on reasoning without external feedback. The useful follow-up confronts the answer with a provided source or with a precise critique from you.

Sources