Last post we looked at how giving AI more context leads to better results.
This week, there’s one small addition to your prompts that can make a
surprising difference — and it takes three extra words.
Ask it to think step by step.
When you ask ChatGPT or Copilot a straightforward question, it gives you a
straightforward answer. That’s fine for simple tasks. But for anything
involving reasoning — working through a problem, making a recommendation,
writing something with a logical structure — the AI can rush to a conclusion
just like a person might when put on the spot.
Why Does This Work?
Adding “think step by step” or “explain your reasoning” is like asking a
colleague to slow down and walk you through their thinking rather than just
handing you an answer. You get to see how they got there, spot anything that
doesn’t add up, and end up with a result you actually understand and trust.
It sounds almost too simple. It works anyway.
When to Use It
This isn’t something you need every time. It earns its place when:
- You need the reasoning, not just the answer. Asking AI to help you decide
between two options? Ask it to weigh them up step by step rather than just
pick one. - The task has multiple moving parts. Anything involving sequencing, planning,
or structured thinking benefits from this approach. - You want to catch mistakes. When the AI shows its working, errors are much
easier to spot than when it just hands you a conclusion. - You’re explaining something to someone else. If you need to justify a
decision or communicate a process, a step-by-step breakdown gives you the
structure ready-made.
3 Office Examples
Scenario 1: Deciding between two options
Instead of: “Should I use Excel or a project management tool for tracking our
team’s tasks?”
Try: “I need to track tasks for a team of 6. We currently use Excel but it’s
getting messy. Compare Excel and a project management tool like Trello or
Asana for our situation. Think through the pros and cons of each step by step
before giving me a recommendation.”
You’ll get a reasoned comparison you can actually present to your manager, not
just a one-line answer.
Scenario 2: Planning something with multiple steps
Instead of: “How do I organise a team away day?”
Try: “Help me plan a one-day team away day for 12 people. Walk me through each
step I need to take, from initial planning through to the day itself. Include
things I might forget.”
The “walk me through each step” instruction does the same job here. You get a
checklist rather than a paragraph.
Scenario 3: Understanding a decision before you commit to it
Instead of: “Is it worth upgrading to Microsoft 365 Business Premium?”
Try: “We’re a small office of 10 people currently on Microsoft 365 Business
Basic. Explain step by step what we’d gain from upgrading to Business Premium,
what it would cost, and whether it’s likely to be worth it for a small
non-technical team.”
Seeing the reasoning laid out makes it much easier to take that to whoever
holds the budget.
A Small Tweak, A Big Difference
You don’t need to use the exact phrase “step by step” — variations work just
as well:
- “Walk me through your reasoning”
- “Explain how you got to that answer”
- “Break this down for me”
- “Take me through this one step at a time” The common thread is that you’re asking the AI to show its work rather than
just deliver a verdict. That 20 extra seconds of typing gives you something
you can follow, question, and build on — rather than just accept. AI works best when it thinks out loud. All you have to do is ask.
Next post:
how to use AI to get more out of the tools you already use every
day — starting with Excel.
Leave a Reply