Can AI Replace Facebook Ad Freelancers?

Are we there yet?

Lead generation Facebook advertising freelancers are probably going to get whacked by AI in the long term.

But exactly how close is AI to taking our jobs away in February 2025?

To see the state of play for AI in Meta ads, we can break it down into four functions:

1. Copywriting

2. Creative Design

3. Media Buying

4. Strategy

In the future, we might have all four mixed into one function plus the functionality of auto-created landing pages. But let’s review each one at a time.

1. Copywriting

Most people have probably played around with using ChatGPT to generate text for something.

And probably also noticed that the generic output sucks.

But at the risk of being a bro, I have to say, 'You didn’t prompt it right.' AI-generated copy can yield great results for immediate use or with some editing.

Here are 6 example levels of copywriting prompting:

1. Generic instructions, “Write an ad for XYZ company”

2. Specific instructions, “to ABC audience, promoting features 1,23”

3. Specific with examples, “… for example {insert from other ads you’ve written}

4. Step by step with examples,

   1. “first, understand this business such that you could write Meta ads copy when promoted. {Business details}.”

   2. “next understand this product and market such that you could write Meta ads copy when prompted. {Product & target details}

   3. “now write an ad for XYZ company… to audience … for example…”

5. Step by step with modular examples, "Write an ad for XYZ company with this framework Headline, Subhead, Product Intro, Product Details, Social Proof and CTA. Here are 3 examples for each element…”

6. Step by step with modular examples AND tweaking, “Now, evaluate the copy from the POV of a {style reference} and re-write based on this.

The output is completely different.

Here’s example ad copy for a Joe’s pizza shop using generic instructions which is where most people start.

Generic prompt (no no tweaking or editing):

Write Facebook ad body copy to promote a pizza shop called "Joe's Pizza".

🔥 Craving the Perfect Slice? 🍕

At Joe’s, we don’t just make pizza—we craft mouthwatering, oven-fresh perfection. Whether you love a classic cheese slice, a loaded meat lover’s pie, or something extra cheesy and gooey, we’ve got you covered.

✅ Hand-tossed, fresh dough daily

✅ Premium, melty cheeses & bold flavors

✅ Baked to crispy, golden perfection

📍 Swing by Joe’s today or order online for pickup & delivery. Because great pizza = great moments.

👉 Tap below to order now! 🍕🚀

Using a level 5 prompt: (no examples, no tweaking or editing):

• Copied and pasted the info from the Joe’s Pizza website

• Copied and pasted probably 50-100 Google reviews + Google maps basic map data

Write Facebook ads body copy for Joe's with this framework call out, product intro, Product Details, Social Proof and CTA targeting locals for the cheese slice with a 50% off discount coupon promotion. For the call out, pick out a quote sentence that could be used as a headline.

🔥 “Joe’s is worth every bit of the hype and the block-long line!” 🍕🔥

Locals know: Nothing beats a classic New York cheese slice from Joe’s Pizza! And for a limited time, we’re treating our NYC neighbors to 50% OFF your next slice!

🍕 What Makes Our Cheese Slice Legendary?

✔️ Crispy, foldable thin crust – the true NYC way

✔️ Perfectly balanced sauce-to-cheese ratio

✔️ Freshly made, straight from the oven

But don’t just take our word for it…

⭐ "The best slice I’ve ever had—crispy, cheesy, and pure perfection!"

⭐ "This is what NYC pizza should taste like. Period."

📍 Exclusive for Locals! Grab your 50% OFF coupon now and enjoy the city’s best slice at half the price!

👉 Claim your discount now! 🍕🎟️

After 1 tweaking prompt (no extra examples):

Critically analyze this copy to see whether it meets the requirements of being non-cheesy, non-try hard from the point of view of a smug Redditor. For example, consider the analogies, tone and wording used. Then, rewrite it based on this analysis.

“It’s just a damn good slice. Nothing fancy, no gimmicks.”

You already know the deal. It’s a proper New York cheese slice, made the way it should be. No overloaded toppings, no weird crusts, no nonsense—just the classic, done right.

And for once, we’re knocking 50% off your next slice—because why not?

Not forever. Just for now. Grab it while it’s here.

👉 Claim your 50% off now.

Not bad. Especially given that I used a basic framework and there’s no editing yet.

Meta is experimenting here too.

When you add your own copy, in some accounts, it will show you it’s own AI generated versions that you can test against your own.

And they’re scarily decent.

Example body copy.

Headlines, too.

Example headlines.

Given that Meta can see all the ads written on their platform and whether they’re working, I wouldn’t be surprised if this means advertisers will eventually just write one version of the copy and then generate the rest on platform.

2. Creative

Images

AI image creatives are far from easily usable, but there is a lot of activity here.

Firstly, Meta is playing with generating creatives itself (see below). They’re pretty cruddy but I think it could work better in eCommerce advertising where people and location context is so crucial to communicating what the ad’s about.

Meta says 11% higher CTRs and 8% higher conversion rate with their generated creatives. Do you believe it?

They allow you to generate backgrounds (presumedly for just eCommerce items) or the entire image. Then add and edit the text overlayed ontop:

Convincing?

Other applications are springing up like Design.com, AdCreative.ai and Predis.ai, but while they’re working hard - so far they’re basically an image generator with a Canva subscription.

Take Predis for example.

It is super cool that it does the copy, images and formats all in one using an AI wizard approach. But the results are pretty underwhelming:

Example creatives generated by Predis.ai for a fitness offer.

The most likely and practical way I’ve seen to make this work is to build your own AI ad generator with a mashup of ChatGPT, Figma and Midjourney:

Video

Surprisingly, it’s the video side of ad creatives that looks like it will get off the ground first.

Creatify has a wizard AI ad generation experience where you put in your company, product and audience details. Then it will generate a script (or you can add your own) with an avatar of your choice.

Here’s an example video ad I made using General Assembly’s digital marketing short course as the sample data:

What’s cool is that you can add your own creative assets to run in the background:

I’m looking to explore this more because with the right script, I think I can make this work.

(Let me know if you’d me to write like a follow up piece about using this tool in the comments).

3. Media buying

Media buying on Facebook ads can be broken down to:

• Targeting

• Budgeting

• Placements

And truthfully, the Meta media buying process has become more and more simplified overtime.

Overall, Meta does offer simplified Tailored campaigns which set up everything for you though not recommend if you consider how they’ve been juicing people using the boosted posts feature.

Budget

The AI automation is already here and prevalent.

A large group of people use Advantage campaign budget which automatically allocates ad spend to targeted groups within your setup that should give you the best performance.

Targeting

For targeting, it’s mostly become a non-consideration too as most people put the bare minimum targeting settings on Facebook ads, filtering just by gender, language, location and age.

A newish feature has been the Advantage+ audience targeting option which relies more on Facebook’s internal targeting parameters, beyond the settings you put in place, to find buyers.

Many get mixed results.

Personally I use it sparingly as the results can be inconsistent. And sometimes you need to set a firm line not to target a certain group of people.

Placements

Placements settings on Meta tell the platform where you want the ads to be shown. For example, on Instagram or Facebook, feeds only or stories and reels.

Facebook has an automation option here called Advantage+ placements which is decent except in certain cases where you can accidentally optimize for poor quality responses.

For example, Advantage+ includes placements on third-party apps and websites:

It’s not consistent whether another website will show your images, headlines or body copy properly (or even at all) which makes it really easy for people to misunderstand your ad.

In one case, I think using the apply now call to action on an add plus the Advantage+ placements may have put my ads onto a job website so people responding thought they were applying for a job. Problem was that the ad was for a paid coaching program!

4. Strategy

Strategy in Meta ad buying includes a few parts:

• Understanding a client’s goals, abilities and resources

• Choosing customer targets

• Researching them to understand desires, fears and interests

• Formulating offers

• Planning funnels

Then mashing them together to create an advertising plan that is effective, profitable and workable for a client.

Predictably, there’s no meaningful progress on this part of the job yet. And I think that’s for a 2 reasons:

• There’s a barrier between the meat-space and cyber-space. Someone has to get that client data on their situation into a machine. AI isn’t quite ready to quiz a client all on their own.

• Inconsistent training materials. Most of Reddit can’t agree on how to run an ad account. So how can AI be trained on that? Even Meta’s own training materials go out of date quickly due to updates in the UI, new features, and the regulatory environment, so they aren’t a reliable source of training material for an AI.

For the adventurous few, I suppose you could try DIY your own part-solution using Chat GPT Projects.

Projects is OpenAI’s tool for housing multiple conversations about the same topic while referencing shared resources and context.

I made my own by adding Meta’s official training materials as PDFs so that I can quiz them on my approach to certain campaign structures.

While it is useful for media buying setup feedback, ultimately it’s not even close to replacing a human with experience and judgment.

---

While AI in February 2025 isn’t close to replacing an experienced Meta media buyer yet, it is making fast progress.

Given the automation in the media component of the job, I think it’s smart to hope your freelancer or agency is leveraging AI to increase their productivity and spend more time making creative, persuasive ads that rock.