10 Prompt Engineering Tips to Improve Your Content Writing and SEO

After a day of work chaos, my therapy doesn’t come in the form of goat yoga or overpriced candles.

It’s cooking.

There’s something about standing over a stove or throwing ingredients together, that calms my brain.

Which leads to a question that I'm always asked: “Can you send me the recipe for that thing you made?”

Unfortunately my reply is always the same: "I don't use recipes."

I’ve been cooking since I was 10.

I make it up as I go, relying on instinct and experience.

But here’s the thing - most people can’t wing it like that.

And if they try, they’ll likely end up with a culinary crime scene instead of dinner.

The same goes for prompt engineering.

Think of the ingredients as data - the raw materials you throw into a generative AI model.

The recipe is your prompt.

A half-baked prompt gets you content no one wants to consume.

A well-crafted prompt is a balanced recipe that turns raw data into high-quality, relevant content.

Whether you’re in the kitchen or crafting AI-driven content marketing strategies, winging it only works if you’ve put in years of practice.

For everyone else, follow the recipe - or in this case, master the art of the prompt.

Because there are still a lot of people who are asking ChatGPT stuff like, “Write me a blog post.”

That’s like asking a chef to reheat airline food.

So here are ten data-driven, borderline-nihilistic prompt engineering tips to make your AI work smarter, not harder (at least that's what I keep telling myself).

1. Be specific

Guess what?

Vague prompts produce vague results.

One of the first funny lessons that I learnt while studying prompt engineering was that if I asked the LLM, "What is a Cloud," it didn't know whether was I was referring to online clouds or atmospheric clouds.

Generative AI relies on patterns from its training data, so if you don’t tell it exactly what you want, you’ll get the most generic and assumptive content imaginable.

Specific prompt engineering ensure that the AI generates relevant, high-quality content.

For example, if I prompt the AI to “Write an engaging H1 for millennial Aussie travellers about budget-friendly destinations in Europe," I give the AI clear boundaries.

The more detail you provide, the less editing you’ll need to do later.

Think of your prompt as a skyscraper being climbed by Tom Cruise: weak prompt = famous actor falling through the clouds.

2. Add context like your SEO ranking depends on it (because it does)

Generative AI doesn’t inherently understand your audience, brand, or goals - it’s not your overworked copywriter who’s memorised every marketing persona.

Let’s say your audience is small business owners.

Telling AI this ensures it uses language and examples that resonate with them.

Skip the context, and your blog might sound like it’s aimed at Fortune 500 CEOs instead.

Example Prompt: “Write a video script for small business owners explaining how local SEO can help them attract more customers. Use a friendly, approachable tone, and include practical examples.”

Context is what transforms generic AI content into something that feels personal, relevant, and useful.

3. Ask for structure and formats

It sends a tear down my cheek to even write this, but online readers don’t actually read - they scan.

Structured content writing with clear headings, bullet points, and short paragraphs is your secret weapon for higher engagement and SEO performance.

I found through my own prompt testing that AI isn’t great at figuring out the ideal structure on its own.

You have to explicitly request it.

For example, if you’re writing a “how-to” guide, specify that you want numbered steps, subheadings, and a conclusion summarising the key takeaways.

Example Prompt: “Write a step-by-step guide on creating an email marketing campaign. Include an introduction, numbered steps with H2 headings, and a conclusion summarising why email marketing matters.”

Without structure for your content, you’re left with a puddle of disconnected ideas that readers - and search engines like Google and Bing - will bounce away from which is a cause for concern.

(I'm just kidding; I don't care what Bing does).

4. Don't shy away from the numbers

Readers trust numbers, and so does Google.

In fact, did you know that Blog posts with credible stats are 58% more likely to get backlinks?

And guess what?

I just made up that stat above.

But you still believed it, didn't you?

Admit it, you believed it...even just a little bit?

With prompt engineering you can ask the AI to include data, but always verify it.

AI loves making up numbers with the confidence of a mediocre poker player.

Example Prompt: “Write a Linkedin post about the benefits of remote work. Include at least three statistics from recent studies on productivity and employee satisfaction.”

Remember: If AI can’t find real data, it will invent plausible-sounding stats.

Use your eyeballs (and Google) to fact-check before publishing.

5. Craft a CTA that actually doesn't sound sales-y

The point of SEO isn’t just to rank - it’s to drive action.

If your content writing doesn’t nudge readers toward something (a purchase, a subscription, or even just a share), it’s like a door-to-door salesman forgetting to close the sale.

Most AI outputs end with bland CTAs like “Check out our website for more information.”

When it comes to prompt engineering, you need to be explicit about the action you want readers to take.

Example Prompt: “Write a social post about why social media is important for small businesses. Conclude with a call-to-action encouraging readers to download our free social media marketing guide.”

6. Optimise with keywords, but don't keyword stuff like 2005

I'm hesistant to suggest this one, but I feel the train has already left the station for this point.

Keywords are still essential for SEO, but overusing them makes your content unreadable and unrankable.

You can prompt AI to weave keywords naturally into the content, but you have to specify what they are and how often to use them.

Example Prompt: “Write a 1,200-word blog post about the best hiking gear for beginners. Use the primary keyword ‘best hiking gear for beginners’ 5–7 times and include related keywords like ‘essential hiking gear’ and ‘beginner hiking checklist.’”

Keywords should enhance your content writing, not dominate it - and I personally would still get a human to check the copy before publishing, and tread carefully relying solely on AI-written long-form content.

7. A/B test your prompts

Different prompts yield different outputs, and finding the best one is part of the game.

Treat AI like a creative partner who never gets tired of brainstorming.

Test variations of your prompts to see which one produces the best result.

Example Prompt A: “Write a Linkedin post about the benefits of mindfulness for productivity.”

Example Prompt B: “Write a Linkedin post about how mindfulness can improve workplace productivity, including one scientific study and tips for beginners.”

A/B testing isn’t just for ads; Apply it to your prompts, and you’ll quickly learn what works best.

8. Remember to stand out from the crowd

The internet is flooded with generic, god-awful content (remember that Game of Thrones Blog, Courtney Fletcher ?)

Powerful prompt engineering involves asking the AI for creative metaphors, unusual perspectives, or even a specific narrative style.

For example, a post comparing SEO to online dating might resonate more than a dry explainer.

Example Prompt: “Write a social post comparing SEO strategies to building the perfect Tinder profile. Use humor and analogies to explain key concepts.”

Memorable content wins every time, so it pays to go a bit above and beyond with your prompts.

9. Tone-of-voice is everything

Your brand’s voice is part of its identity.

Generative AI can mimic tone well, but only if you tell it what you want.

You need to specifiy whether you’re aiming for authoritative, conversational, witty, or even skeptical.

Example Prompt: “Write an email for busy entrepreneurs about productivity hacks. Use a no-nonsense, motivational tone with some light humor.”

10. Ask for proofing and optimisation

Even great AI output needs polishing.

Typos, awkward phrasing, and missed SEO opportunities can hurt your credibility.

Generative AI can help with the final touches, so use it to check grammar, improve flow, and optimise for readability and SEO.

Example Prompt: “Proofread and optimise this blog post for grammar, readability, and SEO. Suggest improvements for clarity and engagement.”

A polished final product is your best shot at impressing readers - and search engines.

Final (important) thoughts

Nail your prompts, and you’re not just creating content; you’re building scalable strategies that outsmart algorithms and uncover opportunities your competitors are too lazy to find.

It’s the difference between playing the game and owning the board.

But a word of caution...

The more we lean into generative AI, the closer we get to a tipping point where content stops being valuable and starts being noise (or AI Content Saturation as I'm calling it).

Imagine a future where every marketing unit is cranking out AI-written blogs like a digital sausage factory.

The value of AI isn’t in how much it can produce; it’s in how intelligently you use it.

If you treat AI like a cheat code for volume, you’re just adding to the landfill.

But if you use it as a tool to elevate your thinking, sharpen your strategies, and free up time to create content that actually connects with people, you win.

Use prompts to enhance, not replace, what makes your content human - creativity, authenticity, and empathy.

Remember, the best SEO isn’t just about gaming the system; it’s about creating value.

And the best content isn’t written by AI; it’s written with it.

Frequently Asked Questions (FAQs)

What is the best way to think of prompt engineering?

Prompt engineering is like being a great bartender.

You’re not just making drinks (or prompts); you’re crafting the perfect mix of ingredients to make something exceptional.

Too vague, and you’ll get watered-down results.

Too complicated, and you confuse the AI like a tourist ordering an "authentic" martini in a dive bar.

The best prompt engineers?

They know exactly what they want and how to ask for it - like the customer who walks in and orders a Negroni, stirred, with a twist of orange, and not a shred of hesitation.

What is an example of using roles in prompt engineering?

Roles are the cheat code to making AI act like it’s got some actual experience.

For instance, instead of saying “Write an article about startups,” you tell the AI: “Act as a venture capitalist with 15 years of experience in Silicon Valley. Write a LinkedIn post about the three biggest mistakes founders make when pitching for funding.”

Now your content doesn’t just sound smart - it has the voice of someone who’s seen enough startup disasters to fill a Netflix series.

Roles turn your AI from a glorified word processor into a convincing expert, even if it’s never paid a single tax.

Which is an example of iteration in prompt engineering?

An iteration in prompt engineering is a cycle of refining and improving a prompt based on the feedback received from the language model's response.

In prompt engineering, I find it is basically the AI version of dating: you rarely get it right on the first try.

You tweak, adjust, and reframe until it feels just right.

Example:

First Attempt: “Write an email marketing campaign for our new app.”

Result: A generic snooze-fest your audience will delete faster than spam from “Prince Olu of Nigeria.”

Second Attempt: “Write a punchy, engaging email campaign for a productivity app targeting busy professionals. Highlight how it saves time and reduces stress, and include a strong call-to-action.”

Result: Much better. You’ve gone from boring to borderline exciting.

Is learning prompt engineering worth it?

Do you enjoy job security?

Then yes.

Learning prompt engineering is like learning to code in the late ‘90s: it might feel optional now, but give it a few years, and you’ll either know it or be working for someone who does.

AI is here, like your new stepfather, and it’s not leaving.

Companies want people who can squeeze every last drop of value out of it.

If you can turn an AI into a content machine, a brainstorming buddy, or even a makeshift therapist, you’ll be indispensable.

How is prompt engineering different from fine-tuning?

Prompt engineering is asking your friend to order a specific pizza for you.

Fine-tuning is teaching them how to make the damn pizza from scratch.

Prompt engineering involves crafting clear, precise instructions to get the AI to deliver exactly what you need, no more, no less.

It’s fast, effective, and requires zero coding skills - perfect for busy marketers, lazy writers, and anyone with a short attention span.

Fine-tuning, on the other hand, means customising the AI itself by feeding it large datasets tailored to your needs.

It’s powerful but also time-consuming, expensive, and about as user-friendly as filing your taxes.

Fun fact: Generative AI doesn't know if the word "filing" means filing your taxes or filing your nails - remember, context is everything!