You Think You're Personalizing. Your Customer Just Feels Harassed.

A few days ago, a friend vented to me.

He'd searched "running shoes" on some app. For the next week, every time he unlocked his phone, a wall of sportswear ads chased him around — same image, same copy, the product color hadn't even been swapped.

He sighed and said: "This kind of personalization? I'd rather they didn't bother."

I stopped for a second when I heard that.

Why?

Because it suddenly hit me — what most brands call "personalized marketing" is basically mass-texting with extra steps. They've just sliced "all customers" into "customers who bought running shoes last time," then sent them all the same email. That's not personalization. That's broadcast with labels.

So what is real personalization?

It's letting every customer walk a path that belongs only to them. What they see next, what they get recommended, which offer finally moves them — all of it decided by their behavior and their preferences, not by your marketing team's content calendar.

It sounds beautiful. It's also genuinely hard.

But since 2023, thanks to AI, it has suddenly become something you might actually pull off.

Today I want to talk about how AI turns the mystic art of "personalized customer journeys" into something you can actually ship.

First, let's get the concept straight: personalized journeys aren't marketing automation

When most people hear "personalization," their first thought is marketing automation. Set the rules: customer signed up three days ago and hasn't ordered? Auto-send a coupon. Cart has items but no checkout? Two hours later, auto-push a reminder.

Sure, that's automation. But it's not a journey.

The core of a journey is that "what happens next" is decided by the customer's behavior — not by a script you wrote in advance.

An analogy.

Marketing automation is like drawing a straight line for the customer — from point A to point B, every step pre-designed, and the customer can only walk along it.

A personalized journey is like handing the customer a park — they can come in through any gate, turn down any path, and what you do is put up a signpost customized just for them at every fork, telling them which way is actually useful for them.

That "signpost" is what the industry calls the Next Best Action — the optimal next move.

And the brain that maps out this park and swaps out the signposts in real time is a class of platform called Customer Journey Orchestration (CJO).

Four roles in this ecosystem — you need to know them apart

I went through the tools on the market, and the platforms that can actually do this fall into roughly four buckets. Let me walk you through them.

Bucket one: CJO platforms. This is what I described above — they string together a customer's behavior across channels (email, web, app, SMS, social), decide in real time what this customer should be nudged with right now, and then execute across all those channels automatically. The customer is the protagonist; the platform is the director.

Bucket two: Real-Time Interaction Management (RTIM) platforms. Very similar to CJO, but the starting point isn't "where is the customer" — it's "I have something important to tell the customer right now; is this a good window for it?" It's more business-driven.

Bucket three: "Next Best Action" platforms. They do exactly one thing: based on a customer's preferences, recent purchases, and browsing history, they use AI to compute an optimal recommendation — an action? a product? an offer? — and then hand that recommendation off to whichever channel executes it.

Bucket four: Personalized images and video. This is the newest and the most imaginative bucket. Combine generative AI with personalized journeys, and a brand can generate different images and video for different customers — same product page, but Alice sees an outdoor scene while Bob sees a street-style city shot.

None of these four works in isolation.

They all need to be fed. Fed what? Customer data. So behind them you need a CDP (Customer Data Platform), a CRM, purchase history, all of it. And you need connected channels — email, web, app, SMS, social — channels that act like tentacles, carrying the brain's instructions to the customer's screen.

That's the ecosystem: a brain + data + tentacles.

Why has this suddenly become non-negotiable?

Three reasons. Each one more concrete than the last.

First: customers are already used to it.

Don't kid yourself that "personalization" is some fancy feature. For today's consumers, it's the default. They open Douyin (the Chinese version of TikTok) and see nothing but what they love. They open Taobao and the homepage is showing what they searched yesterday. They've been trained by these products into an expectation: you should get me.

Whoever gets them, gets their business. It's that direct.

Second: customers jump between devices.

One customer might browse your product on the subway in the morning on their phone, check reviews on their work laptop at noon, and place the order on a tablet at home that night. If a brand can catch them seamlessly across all those devices, they'll be loyal in ways that'll surprise you.

Flip it around: if they add something to cart on their phone and then have to hunt for it again on their laptop — most of the time, they're gone.

Third: the tools have actually matured.

Three years ago, if you'd brought up CJO to me, I'd have said "great concept, brutal to implement." Not anymore. Several mainstream platforms on the market can now hold enterprise-scale volume and actually run in production.

So this isn't a question of "should we." It's a question of "if we don't, we get left behind."

So — where do you start?

Honestly: don't walk in trying to build an omnichannel, end-to-end, fully orchestrated perfect journey on day one. 90% of brands can't pull that off, and even the ones who can, can't do it well.

Here's a four-rung ladder. Climb from the bottom.

Rung One: Nail your automated email first.

If you haven't even stood up behavior-triggered emails, stay away from CJO. Build the drip campaigns driven by customer behavior — what they clicked, what they viewed, what they added to cart, triggering the matching email. That's the fundamentals.

Rung Two: Slip a little more personalization into routine comms.

Even bills, update notifications, everyday pushes — you can slip in a little content that's specific to this customer. One line, one recommendation, one name — a hundred times better than doing nothing.

Rung Three: Put conversational AI to work.

Chatbots aren't stuck at their 2018 idiot-level anymore. Good conversational AI can, while a customer is asking a question, recommend the right product based on their browsing history, and speed up the whole purchase experience.

Rung Four: Run a small, scoped CJO pilot on one or two channels.

Don't get greedy. Pick the one or two channels your customers touch most, plug CJO in, run it for a while, and see what happens. Take what you learned, then expand.

The whole key in one line: ship small, learn as you go.

But — three things you have to watch out for from day one

Which three? Let me take them one at a time.

Number one: data silos.

This is the easiest place to die — right at the start. Marketing has its data. Customer service has its data. IT has its own system. If you want an omnichannel personalized journey and these datasets aren't unified, everything else is empty talk. So from day one, treat "unify the data" as the single most important thing on your plate.

Number two: biting off more than you can chew.

A lot of people walk in wanting to cover every customer, every channel, the entire lifecycle. Result: data isn't wired right, models aren't trained, the team hasn't gelled, and every recommendation that goes out is wrong. The customer came in expecting "you get me," and what they got was "you're worse than a stranger" — and that backlash is uglier than not doing it at all.

So: small scope first, learn first, expand later.

Number three — and this is the most important one — don't make decisions for the customer.

This is the easiest trap to fall into when you're building personalized journeys. You want so badly to "direct," to "orchestrate," that you forget: the final choice belongs in the customer's hands.

There's a delicate balance between what AI suggests, what the brand wants, and what the customer chooses. The best state is: AI gives the customer a few great options, the brand supplies the right information, and the customer makes the call.

The customer should be able to "exit" the current journey at any moment and jump onto a different path. Customers who keep their right to choose are the ones whose loyalty you keep.

A closing note

After my friend vented that day, I went home and dug up some numbers.

The numbers told me one thing: fewer than 10% of brands have actually cracked personalized journeys end-to-end. The other 90% either haven't started, or are still doing pseudo-personalization.

Which is to say: this battlefield's doors are still open.

Whoever first weaves AI into every step of the customer's journey will pick up all those customers who are still getting harassed by broadcast-with-labels.

Maybe one day, my friend will unlock his phone and see — instead of the same running-shoes ad on loop — a video made just for him, explaining exactly why this pair fits his foot type.

At that moment, he won't be wowed by your tech. He'll just feel:

"This brand really gets me."

And here's hoping you become, sooner rather than later, the brand your customers think of as "the one that gets me."