Stop Asking Whether AI Can Do Marketing — Someone's Already Making a Fortune With It

A while back, a friend of mine in the restaurant business came over for tea.

He looked miserable.

Business was slow, he said. He wanted to shoot some short videos, but he couldn't afford an agency, and he didn't know how to film them himself.

I asked him: Have you tried using AI to write a script?

He froze for a second. AI? That thing can write scripts?

It can.

And someone used it to take a 46-second video to 22 million views.

How a Little Shop Selling Tamales Hit 22 Million Views

Let me tell you a story first.

There's a family-run shop in Los Angeles that sells tamales — a classic Mexican dish. It's called The Original Tamale Company. The kind of mom-and-pop operation where the whole family runs one tiny storefront.

Their marketing manager, Christian Ortega, was scrolling one day when he noticed a meme-video trend going around.

He asked himself one question: How can I use this for my business?

So he used ChatGPT to write a voiceover script, edited the video in ten minutes — a guy falls out of the sky and splats right in front of their shop.

Funny, unexpected, and tightly connected to the food they sell.

Three weeks. 22 million views. 1.2 million likes.

Let me be clear — this wasn't really AI's win.

Christian himself said it: if I see an idea that I think will resonate with my audience, I act on it immediately. The key question is always — how does this serve my business?

What did AI actually do here?

Speed.

It let Christian get the video out before the trend went cold. But what actually determined the outcome was that he recognized the meme, judged it as a fit for his shop, and added his own sense of humor.

That's human work. AI just helps human work move faster.

1,600 Designers and Still Not Enough — What Do You Do?

You'd think a company like IBM would have everything it needs for marketing.

They have a design team of 1,600 people.

1,600.

And it's still not enough.

Why?

Because IBM runs ads globally, across different industries and different regions. One creative concept has to spin off into hundreds of variations. Even 1,600 people can't keep up with that volume.

So they partnered with Adobe and used Firefly — Adobe's AI image tool — to generate more than 200 original images, with over 1,000 variations.

The result?

An engagement rate 26 times higher than their previous comparable ads.

And of the people who engaged, 20% were C-level executives — the people who can actually sign off on deals.

Think about who that's most instructive for.

It's not IBM.

It's small teams.

IBM has 1,600 designers. You have three. But AI lets your three people do the work that used to require an entire ad agency. It produces volume you never could have imagined before.

Think AI Is Coming for Your Job? It's Actually Helping You Keep Customers

Let me tell you another one.

Verizon, the American telecom carrier. In 2024, they rolled out several AI systems.

One of them predicted the intent behind 80% of inbound calls.

Plainly put: before the customer's call even connected, AI already had a pretty good idea of why they were calling. It routed them straight to the right person.

The result?

Each customer spent an average of 7 minutes less per interaction. That one system alone stopped roughly 100,000 customers from jumping ship.

But here's the interesting part —

Verizon didn't use AI to lay off its customer service reps.

Quite the opposite. They gave AI to their reps as a weapon. The moment a rep picked up the phone, they already knew what the customer wanted and could offer the right solution right away.

What does the customer feel?

"These people get me."

See, the moment most people hear "AI," their first instinct is — can it save me a few headcount.

That's the wrong way to think about it.

The right play is: AI makes every person you already have stronger.

A Customer Uploads a Selfie, and Conversion Nearly Quadruples

A.S. Watson — the parent company of Watsons and the world's largest beauty retailer.

They did something like this — they built an AI skin advisor online.

Customers come in, fill out a questionnaire, upload a selfie. The AI uses computer vision to analyze more than 14 skin metrics: skin type, skin tone, texture, pores — nothing left out.

Then it generates a personalized skincare plan for you and recommends products along the way.

Guess how the conversion rate did.

Customers who used the AI advisor had a conversion rate 396% higher than those who didn't.

Their average order value was 29% higher.

Why?

Because that in-store counter experience — where a consultant looks at your skin and says, "Here's a toner and moisturizer I'd recommend for you" — got moved online. Instant, scalable, and never tired.

The customer's takeaway: This shop actually understands my skin.

A CEO's LinkedIn Following Grows From 7,000 to 11,000

I especially want to tell this one, because so many bosses agonize over this.

Vector is a company that makes graphics software. Their CEO, Joshua Perk, wanted to build a presence on LinkedIn. But as you know, CEOs are all busy. When they write themselves, quality is inconsistent. When they get someone else to ghostwrite, it doesn't sound like them.

Their head of marketing, Jess Cook, set up a system like this —

First, feed the AI.

Feed it what? Everything the CEO had ever written. His favorite words, his opening habits, the way he used examples — all of it went in.

Then, every week, she'd do a 15-to-30-minute interview with the CEO. She'd ask: What surprised you on customer calls this week? What's a misconception you keep hearing? Tell me about a challenge you just ran into.

Record it, transcribe it, feed it to the AI.

The AI produces 3 to 5 drafts.

A human spends 15 minutes editing. The CEO gives final approval.

The CEO's LinkedIn following grew from 7,000 to 11,000. Inbound customers coming to them to look at the product quadrupled.

Jess said something I really love:

AI gets me 80% of the way there, and then I spend 15 minutes polishing it to 99%.

Here's the thing — AI didn't speak for the CEO. It just learned his voice and turned his real thinking into publishable words.

The soul was still the CEO's.

What Does Ketchup Look Like When DALL-E Draws It?

Here's one that goes even further.

Heinz. The ketchup giant.

They did one thing — they used DALL-E 2 (a text-to-image AI) to generate images of ketchup.

The prompts they fed the AI were "ketchup street art," "ketchup in outer space," "Renaissance ketchup bottle."

Notice — they never mentioned the name Heinz.

But nine times out of ten, the images the AI generated looked like Heinz bottles. That shape, that label, that red.

What does that tell you?

In the AI's mind, the word "ketchup" and the company "Heinz" are bound together.

Heinz turned this experiment into a marketing campaign. They invited the public to generate their own images, then turned selected AI images into real ads and real bottles.

It earned 850 million impressions. Social media engagement was 38% higher than their previous ads. ROI: 1 to 25.

My first reaction after reading this case —

You should immediately use AI to test how tightly your brand is bound to your category.

Ask AI to "draw a project management software." If the AI gives you something vague — an image you can't pin to any specific company — that's not the AI's problem.

It's your brand positioning's problem.

The Little Matter of Email Subject Lines — AI Can Make You Millions

Virgin Holidays.

They used to send out emails, and their in-house or outsourced copywriters could only produce 2 to 3 subject-line options. Slow, mediocre results.

Then they started using an AI copywriting platform called Phrasee, trained on their own brand voice. The AI generated email subject lines automatically and continuously optimized itself based on open rates after every send.

The result —

Open rates went up 2%.

Two percent doesn't sound like much?

For a company the size of Virgin Holidays, that 2% translated into millions of dollars in revenue.

What's the most painful part of email marketing?

You think it's that your writing isn't good enough.

It's actually — you produce 3 subject lines, you can't even test them all properly, so you just pick one on a gut feeling.

AI lets you produce 20 subject lines at once, run A/B tests, look at the data, feed the winners back in. Once that flywheel starts spinning, it only spins faster.

An Intent-Driven Email Flow Lifted Conversions by 82%

One last one — HubSpot.

HubSpot used to send emails by segmenting customers into buckets — "this is a marketing lead," "this is a sales lead" — and blast them.

Then they changed their approach.

They used AI to predict each person's intent. Not which group you'd been sorted into — what you were actually trying to do this time. Then they pushed content that matched your intent.

Conversion rates went up 82% compared to before.

Open rates up 30%. Click-through rates up 50%.

Think about it — going from "I'm guessing you're in marketing" to "I know you're trying to solve an email marketing problem right now." That level of granularity was impossible to achieve by hand.

But now, with AI, it's possible.

What's the Underlying Logic Across All These Cases?

I've covered a lot of ground, and you might be wondering — these companies are wildly different, varying in size and industry. Is there one thing they share?

Yes.

I've thought about it, and it comes down to one thing — AI makes people who already have expertise stronger. It doesn't replace expertise. It amplifies it.

The marketing manager at that tamales shop could recognize a meme and judge whether it was a fit — AI just let him move faster. IBM's 1,600 designers had the taste and the skill — AI let them produce more volume. Verizon's customer service reps had the business knowledge — AI gave them information like a cheat code. Vector's CEO had the ideas — AI helped him get them into writing.

Not a single one of these companies won because "we use AI."

They won because "we have skilled people, and then we gave them AI."

So if you ask me whether you should adopt AI right now,

I'll turn it back on you:

Is your expertise where it needs to be?

If it is, AI will make you move incredibly fast.

If it isn't, AI will only expose your problems faster.

By 2030, There Will Be a More Brutal Shift

One last thing.

By 2027, some predict that mobile app usage will drop by 25%. Why?

Because people won't be tapping on apps anymore.

They'll say something to their AI assistant — "book me a flight," "compare these two projectors for me" — and the AI will handle it.

What does that mean?

It means, if your brand can't be found, understood, or recommended by AI, you "disappear."

Not "ranked on page two of search results" disappeared. Completely disappeared.

In the SEO era, you optimized for search engines.

Next, you'll be optimizing for AI's answers.

Whoever can get ChatGPT, Claude, and Google's AI to recommend their product when relevant questions are asked — wins.

This isn't something that starts in 2030.

It's starting now.

I don't have the answer. But it's worth thinking about — right now.