chatgpt-follow-up-prompts
Module 5 · Lesson 3 of 4 · Course Lesson 18 of 26
Improve Results with Follow-up Prompts
Good prompting is a conversation. Point to the exact problem, add the missing context and request one purposeful change—then verify that the revision actually improved the result.
- Beginner
- About 20 minutes
- Three-step prompt-chain worksheet
- 4-question check
Learning Objectives
What you will learn
By the end of this lesson, you will be able to:
- Treat prompting as an iterative conversation rather than a one-shot command.
- Point to the exact sentence, section, assumption or omission that needs attention.
- Add missing context without unnecessarily restarting the task.
- Use follow-ups to correct, shorten, expand, simplify, reformat and verify a response.
- Ask ChatGPT to identify assumptions, gaps and useful clarifying questions.
- Show a clear quality improvement through a three-step prompt chain.
The Improvement Loop
Review before you request a change
Read
Understand what ChatGPT produced before reacting.
Locate
Identify the exact strength, error, gap or unsuitable part.
Refine
Request one clear change and add missing context.
Verify
Check that facts, meaning and useful parts were preserved.
You do not need to rewrite the full prompt
When the task and source material remain the same, continue in the current chat. Refer to the earlier answer and state what should change, what should stay and why.
Follow-up Prompt Menu
Choose the change you actually need
Correct
Replace a factual error or unsuitable assumption.
Correct the date in paragraph two to 18 August. Keep everything else unchanged.Shorten
Reduce length while preserving essential meaning.
Shorten this to 120 words. Keep all dates, actions and safety warnings.Expand
Add useful detail to a named part.
Expand step three with a beginner example and one common mistake.Simplify
Lower complexity without removing important facts.
Rewrite the second section in plain language for adult beginners.Reformat
Change presentation while keeping the content.
Turn the action points into a checklist. Do not change the wording of the safety warning.Find gaps
Ask what is missing before requesting a revision.
List the assumptions and missing details that could change this recommendation.Ask questions
Let ChatGPT gather context before revising.
Ask me up to four clarifying questions, one at a time, before revising.Verify
Separate supported information from uncertainty.
Identify every factual claim, state its source and flag anything you cannot verify.Annotated Conversation
Three follow-ups, three specific improvements
The learner continues the same task and changes one dimension at a time.
Starting prompt
I run a beginner phone-photography workshop. Write a friendly email reminding registered learners about Saturday’s session from 10 am to 1 pm. Ask them to bring a charged phone and rain jacket.
First response
ChatGPT drafts an email, but the introduction is too long, it invents a venue and it does not ask learners to confirm attendance.
Follow-up 1 · Correct
Remove the venue because I did not provide one. Do not replace it with another location. Keep the confirmed date and time unchanged.
Follow-up 2 · Add context
Add this confirmed requirement: learners must reply by Thursday to confirm attendance. Put it in the final paragraph as a clear action.
Follow-up 3 · Refine style
Shorten the introduction to one sentence and keep the complete email under 130 words. Preserve every confirmed fact and use New Zealand English.
Continue or Start Fresh?
Keep context only when it helps
| Situation | Best choice | Reason |
|---|---|---|
| Same document, goal and source material | Continue the chat | ChatGPT can use the existing response and your earlier requirements. |
| Adding one missing fact or constraint | Continue the chat | A precise follow-up is faster and preserves useful work. |
| Changing format or tone for the same audience | Continue the chat | You can state what should change and what must remain. |
| Completely different task or audience | Start a new chat | Old context may distract from the new goal. |
| The conversation is confused by several abandoned directions | Start fresh or restate a clean brief | A concise current brief can remove conflicting history. |
| Sensitive task should not share earlier context | Use an appropriate separate or Temporary Chat | Apply your privacy policy and minimise unnecessary context. |
Interactive Follow-up Card
Build a precise revision request
Choose one main change. Your entries are saved in this browser.
Prompt-Chain Worksheet
Record three purposeful follow-ups
Write the instruction you used at each stage. Your notes are saved locally in this browser.
Weak Follow-ups
Avoid vague or destructive revisions
“Make it better”
ChatGPT must guess what quality means. Name the exact target and improvement.
Changing everything at once
Separate accuracy, structure and tone changes so you can see which instruction worked.
Missing preservation rule
State which facts, wording or sections must stay unchanged.
Correcting with another guess
Provide a confirmed replacement fact or ask ChatGPT to remove the unsupported claim.
Restarting without need
A new chat discards useful context. Continue when the task and source remain the same.
Trusting self-verification
Asking ChatGPT to check itself supports review, but important claims still need reliable external evidence.
Accepting unintended changes
Compare the revision with the earlier version to ensure good content was not lost.
Endless polishing
Stop when the output meets its purpose, accuracy checks and audience needs.
Refinement does not guarantee truth
A revised answer can sound more convincing while remaining wrong. Open sources, inspect calculations and confirm high-impact information independently.
Practical Activity
Improve one response through three follow-ups
Use a low-risk task and keep the complete prompt chain as evidence. Progress is stored in this browser.
Knowledge Check
Check your understanding
Finish the Lesson
Record your progress
Official References