Marketer, How Long Will Your Job Last?

Not long ago, a friend of mine in consumer goods came to me for tea.

He gave a bitter smile and told me that last year his company ran a marketing campaign — and just the copywriting polish, the creative asset design, the customer segmentation alone meant three months of back-and-forth before anything went live. And once it did? The click-through rate was dismal.

This year?

He had his team try something new. Two weeks, and the campaign was live. Personalization reached nearly every customer. Email click-through rate jumped by a quarter.

Three months became two weeks.

I set my teacup down and asked him: how did you pull that off?

Two words, he said.

AI.

---

Let's Run the Numbers First

Let's zoom out for a second.

One research report put a number on it: generative AI could contribute up to $4.4 trillion a year to global productivity.

$4.4 trillion. What does that even mean? It's the GDP of a mid-sized country.

Of that, marketing and sales grab the biggest slice — roughly three-quarters of the pie. Marketing alone, just that one function, stands to gain around $463 billion a year from the efficiency lift.

$463 billion.

Think about it: that much money is sitting on the table. Whoever gets there first wins.

---

But Here's the Real Question: Where Do You Start?

When most people hear "AI," their first instinct is: this stuff is too out there — let me wait and see what everyone else tries first.

Let me tell you a story.

There's a US craft-supply retail chain called Michaels. They used to send customers emails where only about 20% were "personalized" — which is to say, mass sends, with maybe a name swap at best.

Then they built an AI-powered content-generation platform. And the result?

The personalization rate shot from 20% to 95%. Click-through rate on SMS campaigns jumped 41%. Email, up 25%.

Let that sink in. Same emails. Same customers. Just letting AI help you say the "right thing" to the "right person" — and the gap is that big.

AI isn't here to replace what you do. It's here to help you do what you couldn't do before.

---

It's Not Just About Email

Let me tell you a few more stories.

There's a personal-styling service that has AI help its stylists interpret client feedback and recommend looks. The number of clients a single stylist can serve ballooned several-fold.

There's a toy company that uses AI for product concept design. An industrial designer used to spend seven to ten days drawing a single high-fidelity concept image. Now? Thirty in one day.

Thirty. One day.

There's a food company that uses AI to scan recipes trending on social media, spots which ones could pair with their cereal, and immediately fires off a relevant social-marketing push. The gap between spotting a trend and deploying a campaign has nearly vanished.

What are these companies doing? They're picking up off-the-shelf AI tools and grabbing the lowest-hanging fruit first.

No building their own models. No standing up complicated systems. Buy what's already on the shelf, plug it into existing workflows, and watch the results land.

That's Stage One.

---

Stage Two: Start "Growing Your Own Muscle"

Once you get comfortable with off-the-shelf tools, you run into a problem.

Everyone is using the same set of tools. Your edge over the competition slowly flattens out again.

So what do you do?

A European telecom company's approach is worth a closer look.

They used to send marketing messages sorted into four broad categories. Every customer in the country, slotted into one of four boxes. The problem? That country has several distinct dialects. Send everyone the same voice, and a lot of people simply don't buy it. Conversion rates were brutal.

So what did they do?

They used AI to slice their customers into 150 micro-segments. For each segment, they wrote copy and picked imagery using that group's own dialect, their region's habits, the things their age cohort actually cares about.

The result: response rates rose 40%, and campaign costs dropped by a quarter.

40%. What does that mean? Where you used to spend $100 to acquire one customer, that same $100 now brings in nearly one and a half.

When you use your own data to retrain AI, that's when it starts growing muscle for you. It's no longer everyone's AI. It's yours alone.

---

Here's a Harder One

An Asian beverage company wanted to break into the European market.

The old way: just coming up with a single new product concept would take a year. A full year. By then, the market window could have already slammed shut.

They used AI to do two things.

First, they fed AI a pile of European consumer data and had it analyze flavor trends. A task that used to take a week of market research? AI knocked it out in a day.

Second, they used a text-to-image tool to generate product concept images directly. One day. Thirty high-fidelity beverage concept images. Packaging, flavor, bottle design — all of it. They took those images to real users for testing, and the feedback came back strikingly authentic, because the images looked like real products.

A year's worth of work, done in a month.

Ask yourself: when your competitor can test 30 new products in a month, and you're still spending a year to birth a single one — what exactly are you going to fight them with?

---

Stage Three: Rethink "Marketing" from Scratch

Stages One and Two are still optimizations inside the existing framework.

But some companies are already asking a bigger question: if AI touches every single step of marketing, what does the whole department look like?

Copywriting? AI drafts, humans edit. Market research? AI pulls the signal out of mountains of data. Customer service? AI responds in real time, escalates the hard cases to humans. Campaign planning? AI generates tens of thousands of personalized versions of marketing content in parallel.

This isn't science fiction. The cases above already prove each one can be done on its own. The only question is whether you have the nerve to string them all together.

Of course, this path has landmines.

AI will make things up with a straight face — that's called "hallucination." It can carry bias. It can leak privacy. It can infringe copyright. High-stakes decisions, heavily regulated industries, and scenarios that demand heavy numerical reasoning — these aren't yet ready to be handed over to AI entirely.

So setting up a human-review checkpoint matters more than almost anything else. Anything customer-facing: a human lays eyes on it before it goes out.

---

So How Do You Actually Move?

If you ask me, I'd say take it in three steps.

First, set the direction. Don't try to boil the ocean. Pick two or three of your most painful points and get them running on off-the-shelf tools. Watch where you see results fast and where you actually learn something — then lean into those. Spraying effort across a dozen AI projects means none of them goes deep.

Second, build the team. Three layers. Up top: a coordination office that owns overall strategy and orchestration. In the middle: several cross-functional squads, each owning a specific project. At the base: a technical-foundation team to keep the platform stable and the data secure.

Third, chase quick wins. First six weeks: lock in the roadmap. First ninety days: stand up a "war room" and get your priority projects moving. First six months: run a retrospective, tune the models, and wire AI into your existing marketing systems.

The timeline is tight.

But you already know — this is the kind of thing you can't afford to wait on.

---

That friend I was having tea with later told me something that's stuck with me.

He said he used to think of AI as a tool. After using it, he realized AI is a mirror. It reflected back at him how many things they used to think they "couldn't do" — when really, they were just too lazy to do them, too drained to do them, or too convinced it wasn't worth the effort.

AI has torn down every one of those barriers now.

There's only one question left:

Are you going to act?