The same model gives a far better answer when you ask well.
You don't need tricks or "magic words." You need to give the model what a new colleague would need: context, a clear task, and the shape of the answer you want.
Same goal, two prompts
The 6 habits that do 90% of the work
Give a role
"You are a patient tax advisor…" sets the tone, vocabulary, and depth instantly.
Add context
Who's it for, what's the goal, any background. It can't read your mind — only your words.
Say the format
"In 5 bullet points," "as a table," "under 100 words." You'll get exactly that shape.
Show an example
Paste one example of what "good" looks like. One sample beats a paragraph of instructions.
Iterate, don't restart
"Make it shorter," "warmer," "add a stat." It's a conversation — refine in steps.
Ask it to think
"Think step by step, then answer." Helps on tricky tasks. (Newer "reasoning" models already do this on their own.)
Prompt it well and it gives great answers — but notice it still only ever talks back. The last step is wiring that talent up so it can actually do things for you.