Prompt Engineering Is Just Good Communication

Prompt engineering is just good communication

We gave an old skill a shiny label. "Prompt engineering" sounds technical, but it's the same thing communicators do every day: set intent, give context, and direct the message so it lands.

Large language models respond to clarity the way spokespeople do. When you brief well, you get better output. Communicators are already built for this.

Think like a media trainer, not a coder

Start with the headline in mind. What's the one outcome you want? Define role, audience, format, and tone up front - that's your brief.

Example: "You are a corporate communications writer. Write a 400-word press release announcing our new sustainability initiative for business media. Highlight our 30% carbon reduction target and include a CEO quote on accountability. Tone: confident, plain spoken."

Why it works: you've given purpose, audience, and voice - the same basics you'd give an executive before an interview.

Context is the secret ingredient

Dropping a model into a task without background invites fluff or misses. Add guardrails and signal what success looks like.

• Scenario: Prepare employee talking points for a town hall introducing our new project management platform.
• Context: Rollout starts next month; training is phased; the old tool remains during transition.
• Focus: Simplicity, support resources, why this improves collaboration.
• Avoid: Vendor jargon or long feature lists.
• Format: Five bullets, one sentence each, end with a link to training.

That structure reduces confusion, keeps tone on brand, and saves cleanup later.

Iteration is the new editing

Don't accept the first draft. Treat the model like a teammate: review, redirect, and tighten.

• Round 1: "Summarize this three-page strategy memo for employees in 200 words."
• Round 2: "Make it more conversational and emphasize what is new for teams."
• Round 3: "Open with why it matters, and close with where to send feedback."

Three short passes usually beat one long prompt. The habit of refining is the advantage.

Lead with empathy - the human still matters

AI can draft words. You ensure they land with care. Especially with change communications, tone carries more weight than details.

Example: "Write a 250-word note from a division leader to employees about an upcoming org change. Audience: team members curious about new reporting lines and committed to the mission. Tone: appreciative, transparent, forward-looking. Emphasize what stays the same (mission, customers, core values) and where to take questions."

You're feeding the model the emotional cues it can't reliably infer on its own.

Quick prompt scaffolds for PR work

• Press release: "You are a PR writer. Draft a 500-word release for trade media. Include a quote from the COO, two customer proof points, and a clear CTA for analysts. Tone: credible, plain English."
• Exec Q&A: "Build a 10-question Q&A for a podcast on our new partnership. Include one tough question and a concise pivot response."
• Crisis holding statement: "Write a 120-word holding statement for reporters. Confirm facts we can share, note what is under review, and commit to an update within 24 hours. Tone: accountable and measured."

Guardrails and good habits

• Ask for sources or citations when facts matter. If none are provided, verify independently.
• Feed the model your style guide and past materials to keep voice and word choice consistent.
• State constraints: audience, length, banned phrases, legal sensitivities.
• Use checklists: "Before final, confirm accuracy, clarity, and alignment with our key messages."
• For prompt patterns and examples, see OpenAI's notes on prompting here.

Different tools, same truths

You already know how to brief, focus, and refine. That's the job - and it's also what produces useful AI output.

The next time the cursor blinks, don't overthink the "engineering." Think like a communicator. Set intent, add context, iterate, and apply empathy. That's the craft.

If you want structured practice and prompt libraries for comms teams, explore these resources: Prompt engineering guides.