AI in Marketing: Hyperpersonalization

I gave a keynote this week at a large B2B marketing event in Germany. The topic: "Hyperpersonalization with AI."

Typically, when marketers think of all the great things AI can do, they often dream of serving content at scale that's personalized and meets every user right where they are. Serving the right content, to the right person, at the right time. Since this is notoriously hard to do, everyone’s now looking at AI.

I believe that when it comes to personalization, AI can often do more harm than good. Today, I'll share two examples of what I mean and how I think AI could truly help us deliver more personalized, richer customer experiences.

Let’s dive in!

The 3 Stages of Personalization

Before we get into the nitty gritty details, let me take a step back and show you where we’re currently at (and how we got here).

Back in 1940, if you walked into a diner and wanted a coffee, you had one choice. Coffee. 10 cents. That's it. Maybe milk and sugar if you were lucky.

Fast forward to today: Walk into any coffee shop and the menu looks similar to this:

If you want a Tall Cloud Macchiato with some Caramel Drizzle (for take-away), there’s no problem with that. An entire menu of options tailored to every possible preference.

Stage 1: Customization (Let users choose)

We call this "Customization" and it’s the first - rather primitive - form of personalization. The customer can pick and choose to tailor the product or service to their preferences. This doesn’t just happen in B2C markets. There are plenty examples in B2B as well, like the Siemens Product Configurator that literally lets you choose from 1 billion combinations to pick the drive solution. that’s just right for you.

The problem with customization is that it puts all the cognitive load on the user. Instead of helping, it often overwhelms. It's the IKEA effect — you did the work, so you convince yourself it's valuable. But deep down, you're just tired.

That’s where the second stage of personalization comes in: Curation

Stage 2: Curation (AI recommends to users)

AI has been quietly working here behind the scenes for over a decade:

• Amazon shows you the right products from millions of options.

• Netflix knows what you want to watch before you do.

• Your social media feed gets tailored to you.

While heavily driven by B2C, curation also works across industries and in B2B markets. I myself already worked 10 years ago on an ML (today we would say AI)-driven personalization project for an exhibitor online shop, where trade show exhibitors received recommendations for additional services based on their individual booth setup — ranging from extra lead generation activities to basic power supply they might have forgotten to order.

AI-powered personalization isn't new. But most companies still treat it like science fiction.

Up to today, getting the “curation” layer right is mostly a data problem – not an algorithmic problem. And that’s where it breaks down: Curation only works if you have a) enough relevant data points to analyze and b) enough content to curate.

How do you get enough content to curate? There are two options

1. You produce the content manually (which is expensive), or

2. You generate the content programmatically, a.k.a Template hell. Take 10 audiences × 5 topics × 20 variants = 1,000 soulless templates that are technically personalized, but feel dead inside

I've lived through both. Neither is really sustainable.

Which brings us to the present day.

Stage 3: Creation (AI creates content for users)

Welcome to 2025, where AI doesn't just filter — it creates content.

Two notable events recently happened in this regard:

First, Meta announced Vibes: a 100% AI-generated video feed, unique to every user. No human creators. Just algorithms producing personalized content (slop?) at scale.

A few days later, OpenAI launched ChatGPT Pulse: a daily hyper-personalized news feed that knows you're traveling to London and sends you a sightseeing list two days before your flight. Or reminds you about pet care tips after you mention getting a dog.

So that’s where we are right now.

Why should you even care about this and what are the implications for your business?

Let’s take a look.

How Hyperpersonalization affects your business

The third stage of personalization – creation – is relevant for your business because the technology to do this is already available to you and your customers are already exposed to AI-generated, hyperpersonalized content (as you saw above).

Here's what that means in practice: Your customers' expectations are changing.

When someone experiences hyper-personalized AI content from Meta or ChatGPT daily, they start expecting similar experiences everywhere. The bar for "good enough" personalization keeps rising.

But here's the trap: Most businesses respond by trying to automate more personalization. More emails. More variants. More content generation.

That's the wrong move.

The 2 Paths Forward

Let me show you two examples of how companies are approaching this — and why one works while the other backfires.

Path 1: The Automation Trap (When More Is Less)

This post got 6x more engagement than the original Vibes announcement by Meta's AI chief Alexandr Wang: "Finally, Pure Slop"

This hit a nerve. Naval Ravikant summed it up perfectly: 

Reactions to ChatGPT Pulse were generally more positive. But skepticism existed. Some examples:

• "It makes up broken code examples relevant to my interests"

• "All the MLB playoff graphics are wrong"

• "ChatGPT Pulse reminded me to eat today"

A recent Bynder study found that 56% of people prefer AI content in blind tests. But 70% lose interest when they know it's AI-generated.

Quality isn't the problem. Trust is.

The automation trap is simple: AI lets you create infinite personalized content at near-zero cost. So companies do exactly that — flooding channels with technically personalized but fundamentally meaningless content.

It's personalized spam at scale.

And this might hurt your brand more than it helps.

Eduardo shared a striking example of this:

You don't want your brand being associated with "personalization" that way. It just reveals that in reality, no one is paying attention. 

So what's the alternative?

Path 2: The Amplification Approach (When Less Is More)

Now let me show you a different approach — one I've been using successfully in many B2B contexts – from prepping salespeople to enhancing personalized outreach campaigns.

Instead of using AI to create more content, I use it to research better context for the one piece of content that actually matters.

For example, I built a simple workflow that goes from cold email address to contextual, relevant message in about 5 minutes:

1. Start with an email address — just a business contact email

2. Company enrichment — AI pulls real-time data (industry, size, key demographics)

3. Market intelligence — Using AI and web search to find specific info relevant to my offering (markets, trade shows, announcements)

4. AI scoring — Evaluate fit based on their actual behavior and stated intentions

By the end, you'll get a hyperpersonalized email message that converts 50% better and passes every AI text detector because it’s so specific...

All running in the background while I'm doing other work.

The key difference: I'm not sending 10,000 messages. I'm using AI to do 2 hours of research in just 2 minutes — so I can send one message that actually matters and add genuine relevance.

More importantly: The AI did the homework. A human still made the decision about whether to send it.

What This Means For Your Marketing Strategy

The technology doesn't care which path you choose.

Path 1: Generate 10,000 personalized emails in 10 minutes. Technically possible. Scalable. Probably spam.

Path 2: Use AI to create the one message that actually matters. Ten seconds of effort instead of an hour of research. With human judgment on relevance.

Most companies will choose Path 1. It's easier. It scales better. It looks good in reports.

And that's exactly why Path 2 will win.

Because while everyone else floods inboxes with AI-generated personalization, companies using AI for amplification instead of automation will stand out.

Hyperpersonalization doesn't always mean more. Not more options. Not more content. Not more emails. Sometimes it means the opposite: Having the courage to strip away everything irrelevant.

Amplified relevance > Automated messaging

Because at the end of the day, nobody wants a Cloud Macchiato with Caramel Drizzle at 6 AM when they just need coffee. They want the right thing, at the right time, for the right reason.

So where does this leave you?

If you're thinking about AI-powered personalization for your business, ask yourself:

Are you using AI to automate more, or to amplify what matters?

You don't win the AI personalization game by sending more messages, but by sending the ones that actually matter. Your leverage:

• Research context faster (not create content faster)

• Filter signal from noise (not amplify the noise)

• Enable human decisions (not replace them)

AI can help you create real, personalized experiences. But only if you're brave enough to use it for clarity instead of chaos.

The choice is yours.

See you next Friday,
Tobias