Sometime in the last eighteen months, a quiet divide opened up inside every office that uses AI. On one side, people who type "write me an email" into ChatGPT and get back something generic and vaguely robotic. On the other side, people who get back polished, useful, nearly-final drafts that sound like they wrote them. Same tool. Same subscription. Completely different results.
The difference is prompting. And right now, most businesses treat it as trivial.
The New Literacy
I think of prompting the same way I think about Excel fluency in the early 2000s. Back then, everyone had Excel on their computer. Most people could open a spreadsheet and type numbers into cells. But the person who understood VLOOKUP, pivot tables, and conditional formatting? They were worth twice as much to any team. Not because they were smarter. Because they could extract more value from a tool everyone already had.
Prompting is in that same moment right now.
What Bad Prompting Looks Like
Here's a real example from a professional services context. Someone at a financial advisory firm needs to draft a follow-up email after a portfolio review meeting.
Bad prompt: "Write a follow-up email after a client meeting about their portfolio."
The output will be generic. It'll mention "reviewing your portfolio" and "discussing your goals" and use phrases like "please don't hesitate to reach out." The advisor will read it, sigh, and rewrite most of it manually.
What Good Prompting Looks Like
Better prompt: "Write a follow-up email from a financial advisor to a long-term client named David Chen after a Q1 portfolio review. Tone: warm, professional, not salesy. We discussed rebalancing his equity allocation from 70/30 to 60/40 given his retirement timeline of 8 years. He was concerned about missing tech sector gains. I reassured him that the shift is gradual and we'd revisit in Q3. Action items: I'm sending an updated Investment Policy Statement by Friday, and he's going to review his insurance coverage. Keep it under 200 words. Sign off as 'Sarah.'"
That prompt takes maybe 90 seconds to write. The output will be usable with minimal editing. The difference has nothing to do with knowing secret tricks. It's about clarity of thought. Good prompting is good thinking, externalized.
The Skill Behind the Skill
The people who prompt well tend to be the people who think clearly about what they want before they ask for it. They know their audience. They can articulate constraints. These are communication skills. They're the same skills that make someone good at delegating to a junior team member.
A study by Harvard Business School and BCG found that consultants using AI completed tasks 25.1% faster and produced 40% higher quality output when given proper guidance on how to interact with the tools.
Teaching It Inside Your Organisation
Most AI training focuses on the tool. Here's how to log in, here's where to click. That's fine for day one. But the real training is in prompting, and it should look more like a writing workshop than a software tutorial.
Run a session where your team brings real tasks. Not hypothetical ones. Actual emails they need to send, actual reports they need to draft. Have them write prompts live, share the outputs, and iterate as a group. I've run sessions like this where the collective "aha" moment is visible.
Some practical patterns: always specify audience and tone. Give length constraints. Include examples of what good output looks like. Tell the AI what to avoid, not just what to include. Break complex tasks into steps rather than asking for everything at once.
If you invest in prompting skills, those skills carry across every AI product. ChatGPT, Claude, Copilot, whatever comes next year. The underlying principle doesn't change: clear input produces better output.
The tool is free, or close to it. The skill is what costs time to develop. Start now.