Why your AI prompts keep failing and the 4-sentence fix that changes everything

You pop something into an AI tool, hit enter, and… well, it's okay. Sort of. Not quite what you actually wanted, so you try again, and again. By the third or fourth prompt, you're wondering if AI is really all it's cracked up to be.

It is. The problem isn’t the tool; it's the conversation you’re having with it. Most people engage with AI as if it were a vending machine: press a button, get a result. Generic output is par for the course because the input was vague.

The solution is simpler than it seems and, once understood, will make an immediate difference.

Why Your AI Keeps disappointing you

Ethan DeWaal, AI Program Manager at Asana, trains teams on AI adoption every week. He hears the same three complaints on repeat:

 “It feels unnatural.” “It's not using our voice.” “I had to rewrite the whole thing anyway.”

DeWaal believes people are prompting AI as if they were yelling across a loud room-without context, background, or framework, just a task thrown into the void with a wish. Think of AI as an eager golden retriever. It desperately wants to help you, but without clear direction, it'll fetch whatever's closest. And that's rarely what you actually need.

The 4-sentence framework that fixes your prompts

DeWaal's system is simple enough to memorize in a minute and powerful enough to transform your output immediately. Every great prompt has four parts:

• Persona: Tell the AI who it is before you tell it what to do. This shapes its tone, vocabulary, and expertise level instantly.

"You are an experienced operations manager who specializes in clear, jargon-free communication."

• Goal: State the bigger objective. Give the AI the "why" behind the task. Context about the end goal produces far more relevant output than task instructions alone.

"Your goal is to help operations leaders share project status without adding meetings to the calendar."

• Task: Specify the desired action. Now give the instruction. With persona and goal already set, even a short task directive produces sharp, focused output.

"Draft a weekly project summary template that takes under 5 minutes to complete."

• Context:  add the details that shape the output. Formatting rules, what to avoid, the intended audience, tone, length, examples. This is where good prompts become great ones.

"Keep it under 200 words. Use plain language. Avoid jargon. The audience is non-technical leadership."

Four sentences. That's the entire system. And the difference in output quality is not subtle.

3 Habits to Ensure Your AI Prompts hit the mark

• Provide a real example. Want AI to mimic your brand's style? Paste a past piece of content that’s been approved and tell the AI, "Match this style." A concrete example beats a lengthy description every time. The AI will reverse-engineer the tone, structure, and rhythm for you.

• Correct it, don't restart. "This sounds too formal, make it more conversational" or "Cut this in half and lead with the result" are perfectly valid next prompts. Refining is faster than starting over, and the AI holds context between turns.

• Use an up-to-date, high-quality model. A subpar AI experience can simply be a result of using an outdated version. The leap in capabilities between last year's and this year's models is substantial. Ensure your team has access to the best tools for the job.

The data behind structured prompting

This isn't anecdotal. A joint study by Boston Consulting Group and Harvard Business School found that professionals using AI with structured prompts saw measurable, significant results across the board:

(lets put a rectangle thta has all the numbers instead; that is 

1. 12% more productive overall

2. 25% faster task completion

3. 40%+ higher quality output (independent review)

That's not a marginal improvement. That's the difference between AI being a nice-to-have and a genuine competitive edge.

The part nobody talks about

“Better prompts amplify your existing systems,  they don't fix broken ones. Teams without clear ownership, defined processes, or aligned workflows will simply produce faster versions of their confusion.”

The teams seeing the greatest results from AI didn't just learn to prompt better. They aligned their workflows first. They got clear on who owns what, what "done" looks like, and how information flows. AI then became an accelerator for a system that already worked, not a band-aid over one that didn't.

Your AI prompts are only as strong as the system behind them. And if you're ready to build that system properly, we'd love to show you how.  Book a Free 30 minutes Strategy call.