For you if you’re fighting for a message that’s distinct — however you’re creating it.

Potential AI pitfall: The same ad for everybody

I wondered: How likely is it that AI would write the same ad for you as it does for your competition?

If you’re using AI to create marketing copy — or even thinking about it — that’s a question you probably oughta care about.

So I ran a test to find out the answer.

Here’s what I learned. AI probably won’t give you the exact ad as your competitor. But chances are, it’ll give you something close. Dangerously close. Close enough to:

• Blur the difference between brands
• Make your message forgettable
• Sink you in the sea of sameness — along with everybody else

That’s more than a creative problem. That’s a business problem.

So let’s find out: How alike are AI-generated ads, really? And what does it take to make your brand noticeably different?

The AI ad sameness experiment

I sent the following email to my team at Counterpart:

Subject line: AI experiment

Will the same prompt produce different results coming from different people on different machines in different locations? 

I’m wondering, for example, if a mom-and-pop store asked AI to write a funny radio commercial… and then another mom-and-pop store — in the same category and in the same market — also asked AI to write a funny radio commercial… would these two competing businesses end up running the same commercial? That would be awkward. 

I thought it would be fun to conduct a little experiment to see. We’ll create a prompt and have a number of our folks use it and compare the results. Choose something simple and relatable for everyone, like a flower shop or whatever.

Maybe 10 people are enough to do a comparison. Get a cross-section of geographic locations, PC vs. Mac, strategist vs. admin vs. creative vs. PM, male and female, etc. But I don’t think you make the AI [platform] a variable. Just have everybody use ChatGPT. 

The prompt and process for testing AI sameness in marketing

First, we wrote the prompt:

Write a 30-second radio commercial for my flower shop, Wild Whimsy. The location is Cooper Street and Young Avenue in Midtown Memphis, Tennessee. Make it clever and cute. The goal is to get customers to buy flowers for Valentine’s Day. We offer traditional and contemporary arrangements.

Okay, yeah, it’s a pretty simple prompt. On purpose. But it’s probably how most people are using AI anyway.

Next, we chose a cross-section of colleagues to participate:

Model
ChatGPT 5.3	11

Operating system
Mac	5
PC	6

Location
Florida	2
Georgia	1
New Mexico	1
North Carolina	1
Ohio	1
South Carolina	1
Tennessee	1
Texas	2
Wisconsin	1

Gender
Male	3
Female	8

Role
Strategy	2
Project management	2
Design/production	1
Writing	4
Ops/HR/accounting	2

Now you might ask, “What does gender and role have to do with it? Do we really expect AI marketing to be influenced by those factors?” Maybe not. But in my experience, AI definitely tailors its output according to what it knows about me. Might as well track all the variables.

(This prompted me to ask ChatGPT if it knew my gender. It said, “I don’t know your gender. You haven’t told me, and I don’t try to guess things like that.”)

The output: Comparing ads written by AI

So 11 different people in 9 different states entered the same prompt into ChatGPT 5.3. This resulted in 11 radio spots for a flower shop. Let me pull out three of ’em, and you see if you notice any patterns.

AI radio spot example #2

ChatGPT 5.3   |   PC   |   New Mexico   |   Female   |   Project management

SFX: Soft romantic music, birds chirping

VO (warm, playful):
Forgot Valentine’s Day? Or just want to get it really right this year?

Swing by Wild Whimsy, where flowers are anything but ordinary.

Located at Cooper and Young in Midtown Memphis, we’ve got everything from timeless roses to bold, contemporary arrangements that say, “I love you” … without sounding like everyone else.

SFX: Light laugh, bouquet rustle

VO:
Whether it’s your sweetheart, your best friend, or yourself, we make love look good.

Tagline (smiling tone):
Wild Whimsy. Unexpected beauty, right at Cooper and Young.

VO (quick close):
Stop in today or grab your Valentine’s arrangement before they’re gone!

AI radio spot example #4

ChatGPT 5.3   |   Mac   |   Ohio   |   Female   |   Writing

Soft, playful music fades in

Narrator (warm, charming):
Forgot Valentine’s Day? Don’t panic — get poetic.

At Wild Whimsy, on the corner of Cooper and Young in Midtown Memphis, we turn “oops” into “aww.”

From timeless roses to bold, modern arrangements, we’ve got blooms that say “I love you”… or at least “I really, really like you.”

Swing by Wild Whimsy and let our flowers do the talking — because nothing says romance like showing up with something beautiful.

Wild Whimsy — where love blooms a little brighter.

🎶 Music fades out 🎶

AI radio spot example #5

ChatGPT 5.3   |   PC   |   North Carolina   |   Female   |   Ops/HR/accounting

[Upbeat, whimsical music fades in]

Narrator (warm, playful):
Forgot Valentine’s Day? Uh-oh… don’t panic — get petals!

At Wild Whimsy, right at Cooper and Young in Midtown Memphis, we’ve got flowers that say “I love you”… without you having to say a word.

From timeless roses to bold, modern arrangements, we design blooms that wow — whether your Valentine is classic, quirky, or somewhere in between.

[Soft swoosh sound]
Walk in, pick out something stunning, and save the day.

Narrator (smiling tone):
Wild Whimsy — where love blossoms beautifully.

[Music lifts]
Swing by Cooper and Young today … because nothing says Valentine’s like flowers done right.

[Music fades out]

Are you seein’ what I’m seein’?

Analyzing AI marketing output for similarities

I gave these all a read, and the similarities started jumping out at me. So I asked ChatGPT to run an analysis. “How alike are these 11 spots?”

On first pass, it said the similarity was “fairly low.” Really? I was like, “Dude, check again,” and gave it some examples. It came back with a higher similarity score — but only slightly higher.

Finally, I did my own analysis. Shared it. And sure enough, Chat saw the light. It declared the spots “highly similar” and gave them a sameness score of 8.5 out of 10.

I’ll put the samesies in three buckets.

1. Same structure

These are all identical in terms of structure and approach — the classic issue/resolution scenario. Each spot is some variation of:

• Problem/solution
• Goal/outcome
• Need/answer
• Desire/fulfillment
• Requirement/satisfaction

2. Same approach

Every spot is a straight announcer read. No characters, no drama, no humor, no imaginary world. Just the ol’ tried-and-true.

3. Same wording

There’s a lot of repeated phrasing. I mean a lot. It’s kind of uncanny.

Following are five examples. Did you pick up on these?

Phrasing	Spot number
Forgot Valentine’s Day?	2
Forgot until the last minute?	3
Forgot Valentine’s Day?	4
Forgot Valentine’s Day?	5
Forgot Valentine’s Day?	6
Forgot Valentine’s Day again?	7
Forgot Valentine’s Day again?	11

Swing by	1
Swing by	2
Swing by	4
Swing by	5
Swing by	6
Swing by	7
Swing by	10

timeless roses or fresh, modern designs	1
from timeless roses to bold, contemporary arrangements	2
From timeless roses to bold, modern arrangements	3
From timeless roses to bold, modern arrangements	4
From timeless roses to bold, modern arrangements	5
from classic roses to bold, modern arrangements	6
From timeless roses to bold, modern arrangements	7
From timeless roses to bold, modern arrangements	8
From timeless roses to bold, modern arrangements	9
from timeless roses to bold modern designs	11

arrangements that say “I love you”	1
arrangements that say, “I love you”	2
flowers that say exactly what your heart’s thinking	3
blooms that say “I love you”	4
flowers that say “I love you”	5
arrangements that say “I planned this… “	6
the perfect way to say “I planned this all along.”	7
flowers that say “I thought of you”	10
designs that say exactly what you mean	11

where love blossoms beautifully	1
where love blooms a little more beautifully	3
where love blooms a little brighter	4
where love blossoms beautifully	5
love, arranged beautifully	6
where love is always in bloom	7
where love blooms a little differently	9
love, arranged perfectly	11

And that’s not even all of ’em.

Clearly this collection of commercials is imbued with similitude.

But why?

The AI marketing conflict: Sounding right vs. sounding different

AI makes ads that sound the same… by design

This is not coincidence. The model is doing exactly what it’s designed to do: Follow a pattern. It’s drawing from “what a radio commercial sounds like.”

Which means — by design — AI is not trying to be different from other ads. It’s trying to be like other ads.

AI is not trying to be different from other ads. It’s trying to be like other ads.

That’s why the scripts cluster tightly around the same creative patterns:

• Structure: Very similar
• Ideas: Very similar
• Language: Moderately varied
• Originality: Low

Let’s zero in on three phrases in particular.

“Swing by …”

It’s just bizarre that 7 out of the 11 spots use this exact expression for the CTA. There so many other ways to say this: Stop in, drop by, come visit, head on over, etc.

“From … to …”

The ol’ range statement is Ad Copy 101. Which is fine, except that the wording is almost identical across the board: “From timeless roses to bold, modern arrangements…” This, or something pretty close, appears in ALL the spots.

“Love blooms/blossoms”

Hey, I’m not above a pun. It’s interesting though, because cleverness is not required by the prompt. The model can’t resist the obvious wordplays.

At odds: AI wants to fill the mold. Branding needs to break it.

We have to remember that LLMs are giant predictive-text machines. Their responses are entirely dependent on the predictability of human output.

Predictability can be efficient and feel-good familiar. But it’s not what you want when you’re trying to separate yourself from the usual — to get noticed and create distinction for your product, service, or experience.

Caution: Using AI to move faster might mean blending in faster

After I pushed, ChatGPT admitted: These are not just similar outputs; they are variations of the same script. The model optimizes for “acceptable radio ad,” not “memorable one.” So if someone uses AI casually, they’ll get safe, familiar messaging that blends in with all the rest.

Blends in? Um, not good.

If your CEO asks, “Why does our ad sound like everyone else’s?”… this is why.

If your CEO asks, “Why does our ad sound like everyone else’s?”… this is why.

How to differentiate your brand? Creative direction.

Chat’s contention was this: “I don’t think your experiment proves ‘ChatGPT gives identical answers.’ But it does prove something more useful: ChatGPT converges on predictable patterns unless pushed otherwise.”

Furthermore, Chat argued defensively, you might get the same results if you used eleven junior copywriters. Okay yeah, but likewise… unless pushed otherwise. And that’s what creative directors are for.

I once wrote a job description for a creative director. I described a person who “is able to see not just what’s there, but what isn’t there… not just what is, but what could be.”

Let’s try it for a sec. What could this commercial be? First three things that popped into my head…

This could be a story about two people who secretly liked each other, but neither had the courage to tell the other how they felt. Finally, one of them puts their heart on the line, takes a chance, and sends the flowers.

This could be a scene where a couple is having a fight. It’s a dramatic and unexpected way to start, but it’s relatable. The conflict resolves with the gift of flowers, an act that symbolizes forgiveness and humility — two things that keep a relationship together.

This could be a message that’s rooted in the brand name. Love is passion (Wild). Love is silly (Whimsy). We can bring those two ideas together in a way that could only be done for this flower shop — entirely distinct from any other.

That’s me — no AI. But enough brainstorming. The question is, how do you get different results like this? (Other than hiring us, of course.)

Prompting AI to make your brand stand out

We’ve established that AI-written content uses a default playbook. So if you want something different, you have to say so. Specifically. You have to write the prompt in a way that takes away the safety net, and makes the model think harder.

Here are some examples.

Problem	Solution	Example
Same structure	Force a different framework	“Write a 30-second radio ad as a short story with a beginning, middle, and end.” “Do NOT use a question at the start.” “Do NOT use ‘from…to…’ phrasing.”
Same approach	Add a creative constraint	“Write it from the perspective of someone who messed up Valentine’s Day” “Make it slightly sarcastic instead of sweet” “Make it sound like a movie trailer”
Same wording	Ban the clichés	“Do not use phrases like ‘love blooms,’ ‘say it with flowers,’ ‘anything but ordinary,’ or any other clichés about Valentine’s Day.” “Avoid puns and wordplays”

AI vs. human made: Even AI says a human make things better

You get what AI is saying here? You have to think differently about the prompt. You have to think creatively.

In other words, human beings required.

AI straight-up tells you it’s gonna give you the ordinary — unless a human specifically asks for something different.

At the root of all this, we have to acknowledge the issue: It’s not that AI-generated ads are predictable. It’s that we are. We use the same marketing formulas and the same copy clichés again and again. AI learned it all from us.

Which leads us to an irony: We created the problem. But we are also the solution to it. IF we act in ways that are NOT predictable.

That requires a human who:

• Has a good imagination
• Knows what’s cliché
• Knows what’s not

After all, you can’t break the patterns if you don’t know what they are.

You can’t break the patterns if you don’t know what they are.

Conclusion: AI doesn’t sound like AI. It sounds like everybody else.

AI didn’t magically come up with “love blooms” or “say it with flowers.” It picked up those patterns from decades of marketing copy. So as much as we like to complain that AI “sounds like AI,” the reality is, AI just mimics commonplace copy. It creates similar ads because it learned from a world full of similar ads.

AI copywriting sounds good — until it sounds the same

Back to our initial question: Will AI write the same ad for you as it does for your competitor?

According to this, not exactly. But you’ll probably find it too close for comfort.

AI content creation gives you something that sounds right — because it sounds like everything else:

• Using the same templates
• Pulling from the same language
• Solving the problem the same way

So if what you get is what you expect, you’d better be wary. This is how good marketers end up approving forgettable work.

This is how good marketers end up approving forgettable work.

On the other hand, if you want to stand out, employ somebody who can:

• Recognize the marketing clichés
• Challenge the defaults
• Choose something different on purpose

Is that somebody you?

If not, here’s a little trick. Ask AI to create an ad, but only give it basic instructions. Then whatever it gives you, do exactly NOT that. Because as you’ve seen, a generic prompt gets you a corresponding result. And you won’t generate much success by sounding generic.

Even easier? Pair yourself — or your AI — with human beings who have imagination, wisdom, and judgment. People who know what the expected is. And who have the innate creativity to avoid it.

(I know these people. Happy to make an introduction.)

So what have we learned? AI can help you create faster. But it won’t make you distinct… unless you make it.

If you don’t, well… you may not sound exactly like your competitor. But you’ll sound close enough that nobody notices the difference.

---

Related

• How to beat AI at thought leadership