Recently, a number made me stop in my tracks.

The global AI marketing market is projected to surpass $100 billion by 2028.

$100 billion.

My first reaction wasn't "wow, that's huge." It was a much more basic question:

Where is all that money actually going? Which brands are using AI to build things that genuinely drive revenue, reach, and retention?

With that question in mind, I dug through the most notable case studies from the past couple of years. After reading them all, one thing stood out:

The best AI marketing never shows off.

Every one of them used AI to solve one specific, razor-sharp problem.

Let me tell you a few stories.

Let's start with an old brand.

Heinz ketchup. Over 150 years of history. There's probably a bottle sitting in your grandmother's kitchen right now.

What does a 150-year-old brand fear most?

Being seen as "outdated" by young people.

In 2022, Heinz did something. They used DALL-E 2 to have AI draw ketchup bottles.

How?

They fed the AI all kinds of outrageous prompts. "A ketchup bottle in Renaissance style." "A cyberpunk ketchup bottle." "A Chinese ink-wash painting ketchup bottle."

Then, something interesting happened.

No matter how you prompted it, no matter what style you switched to, the bottle the AI drew always looked like a Heinz.

Why?

Because the Heinz bottle silhouette is so iconic that in the AI's training data, "ketchup bottle" is practically synonymous with "Heinz."

Brilliant.

Once Heinz noticed this, they turned it into a full-blown marketing campaign. They hosted a metaverse art exhibition, released limited-edition bottles, and invited the internet to play "AI draws ketchup" along with them on social media.

The result?

850 million impressions. Media ROI exceeding 2,500%. Engagement rate 38% higher than their previous campaigns.

Even totally unrelated brands like Ducati and sports TV networks came knocking: "Hey, make us an AI ketchup bottle too."

Think about it — what makes this so clever?

Heinz didn't spend a single cent "educating" consumers that they were the real ketchup. They just let AI say it for them.

When even AI thinks ketchup looks like you, do you still need to run ads?

That's using AI to make "brand equity" tangible.

Turning a Burger Into a Million-Dollar Spectacle

Here's another one.

Burger King. 2024.

They had this old slogan, "Have It Your Way," that had been around for decades. How do you breathe new life into a slogan that old?

Burger King launched a contest. A million-dollar grand prize.

The rules were simple: use their app to build your own Whopper ingredient combination and submit it.

But the contest itself wasn't the point.

The point was this — after you submitted your creation, Burger King used AI to generate a "poster-grade photo" of your Whopper, paired with a custom AI-generated jingle just for you.

Picture this: you spend two minutes topping a burger, and you get back a professional-grade product shot plus a song written for you.

What would you do?

Post it to your feeds. Share it on Instagram. Put it on X.

Every single entry came with built-in virality.

That's the genius of it. Burger King didn't pay anyone to post. They just gave users something they "couldn't resist sharing."

The best marketing isn't getting users to see your ad — it's turning users into your ad.

That's what AI did in this case: it wasn't doing the creative work, it was lowering the barrier to "user participation." In the past, if you wanted users to spread the word, you had to give them a reason. Now AI turns that reason into ready-made content.

Serena Williams From 1999 Plays Serena Williams From 2017

Nike's story is even wilder.

In 2022, Serena Williams was nearing retirement. Nike wanted to honor her.

What would a typical brand do? Cut a retrospective short film, layer on some sentimental music, done.

Nike didn't do that.

They brought in a tech team and used AI and machine learning to stage a virtual match.

What kind of match?

Serena Williams winning her first Grand Slam in 1999 plays Serena Williams winning the Australian Open in 2017.

The 1999 version: 18 years old, just starting out, lightning reflexes, aggressive play.

The 2017 version: 35 years old, seasoned veteran, steady rhythm.

Two eras of the same person, playing each other across time.

AI analyzed enormous amounts of match footage — shot selection, reaction times, movement patterns, body posture — then used that data to build a digital avatar for each version of Serena.

The match was livestreamed on YouTube.

How many people watched?

1.7 million.

Even more staggering: compared to Nike's usual content, organic views on this one jumped 1,082%.

1,082%.

Why does this story move people?

Because Nike didn't use AI to make a "smarter ad." They used AI to do something "that was simply impossible before."

True innovation isn't doing the old things faster — it's making something that has never happened, happen.

An athlete competing against her younger self. That can never happen in the physical world. But AI made it happen.

That's the most captivating thing about AI: it doesn't just save you time and money — it creates an entirely new dimension of experience.

Saving $10 Million a Year, Cutting Production From 6 Weeks to 7 Days

Enough creativity — let's get practical.

Klarna, the Swedish Buy Now, Pay Later (BNPL) company. In 2024, they baked AI into their marketing production pipeline.

How?

It used to take about six weeks to produce a set of marketing assets — from shooting, to retouching, to delivery. Klarna used generative AI to compress that down to seven days.

Six weeks became seven days.

In three months, they produced over 1,000 customized marketing images.

And they sharply reduced their reliance on vendors and external agencies. A lot of it, they just handled in-house with AI.

How much did they save?

About $10 million annualized.

$10 million.

Think about what that means for a company. This isn't money saved "by cutting budgets" — it's money saved "by changing the way they work."

But Klarna's smartest move wasn't about how much they saved.

It was treating AI as a "production system," not a "one-off stunt."

A lot of companies use AI in marketing like this: spend a week on an AI concept, send out a press release, trend for a day, and then... nothing.

Not Klarna. They wove AI into their daily workflow. Producing images, revising them, testing, iterating — all with AI.

AI's real value isn't in that one dazzling moment — it's in the repetitive work, day after day.

One Billion Virtual Try-Ons, Conversion Rate Triples

L'Oréal's story made me go "oh?"

In 2019, they launched two AI tools. One called ModiFace, for virtual try-on; the other called SkinConsult AI, an AI skin diagnostic.

Virtual try-on uses AR technology — you take a selfie with your phone and see how a certain lipstick or foundation looks on you.

The skin diagnostic has you upload a selfie, and the AI analyzes your skin condition: hydration, firmness, wrinkles. Then it recommends products.

Both tools were embedded across L'Oréal's brand websites and apps, as well as retail partner platforms like Amazon.

The numbers came in:

Virtual try-on was used more than one billion times globally.

Users who tried virtual try-on converted at three times the rate of those who didn't.

Three times.

Why?

Because the biggest barrier to buying cosmetics is "uncertainty." Will this lipstick actually suit my skin tone? Will this foundation look unnatural on my face? If you're not sure, you don't buy.

Virtual try-on eliminates that uncertainty.

What's more, the skin diagnostic generated 20 million personalized skin reports. That data, in turn, helped L'Oréal better understand its users and guide product development.

Trust isn't something you declare — it's something that grows naturally once you've solved a problem for your users.

The King of Bollywood Endorses Your Little Shop

The last one made me gasp.

Cadbury. Diwali 2021, in India.

They cast Bollywood superstar Shah Rukh Khan to star in an ad.

But not an ordinary ad.

They used AI and deepfake technology to automatically swap in Khan's face and voice, matched to different small shops' names and categories.

What does that mean?

It means the little candy shop on the corner could get its own video ad of "Shah Rukh Khan endorsing your shop."

A shop owner just went to the campaign website, filled in their shop name and category, and within seconds received a personalized celebrity endorsement video, ready to post straight to social media.

The result: during Diwali, the campaign reached 140 million people. Over 2,500 small shops got their own celebrity endorsement videos. Brand engagement rose 32%.

And the case won a Cannes Lions award.

Why does this story excite me?

Because it took "celebrity endorsement" — something only big brands could afford — and turned it into something small merchants could use too.

AI's greatest power isn't making the strong stronger — it's giving small players access to privileges that used to belong only to the big.

Underneath All These Stories, There Are Only Three Things

I went over these case studies again and again, and realized they're all really just doing three things.

First: turn users into broadcasters.

Heinz let users play with AI drawing ketchup. Burger King let users show off their custom Whoppers. Barbie let users paste their faces into the movie poster (13 million uses, 13 million shares). The common thread? Brands stopped producing ad content themselves and instead gave users a tool to produce and spread it on their own.

Second: push efficiency to the limit.

Klarna compressed image production from six weeks to seven days, saving $10 million a year. Mango used AI to generate visuals for entire product lines. The Washington Post's Heliograf wrote 850 articles in a year. They're all doing the same thing: using AI to replace repetitive labor and free people up for higher-value work.

Third: personalize down to "only for you."

Netflix's recommendation engine drives 80% of what people watch. Spotify's Wrapped year-in-review got 60 million people to share it on their own. L'Oréal's one billion virtual try-ons. Amazon's recommendation system. The logic is the same: when data and AI are powerful enough, the experience you give each user can feel as if it were custom-made for them alone.

You can pick one of these three and think about how to apply it to your business today.

You don't need a $100 billion budget.

You need one specific problem — small enough to be razor-sharp — and then ask yourself: can AI help me do this better?

The answer, most likely, is yes.

I'm also using AI to reshape my own workflow. The biggest thing I've realized in the process is this:

AI won't replace people who are willing to think — but it will leave people who refuse to think with no chance at all.

Here's hoping you become the one who's willing to think.