From mass spam to fake AI personalization

It's not more relevant. It just simulates being so.

This morning I opened an email from someone I don't know. It wasn't the message that stopped me, but the intention I felt behind it.

There was no value, no new idea. Just my own words returned (highlighted in bold), rearranged to seem relevant.

And instead of feeling seen, I felt watched.

Mass spam never pretended to care. It arrived, you ignored it, done. This is different. It doesn't arrive as noise, but as apparent interest. But there's no real interest. There's a simulation built with what you already said, just enough to seem relevant for a moment.

"When personalization has no substance, it doesn't build trust. It erodes it."

Real personalization doesn't replicate. It contributes something that wasn't there before: an external perspective, a non-obvious connection, or a concrete reason why that conversation makes sense with that person and not just anyone.

This didn't start with AI

This isn't new. It's the evolution of a practice that has existed for a long time, only now it has more sophisticated tools to hide behind.

First came raw volume: thousands of identical emails, with no pretense of personalization. The game was statistical, not relational. At least it was honest in its cynicism. It didn't pretend to know you.

Today we see the same on LinkedIn

Sometimes obviously, with messages that say "I saw your podcast" without having listened, or "thanks for visiting my profile" when you never did. Other times more subtly: a connection request with no context that, once accepted, becomes a direct sales pitch.

The problem isn't automating. It's using it as a shortcut.

"A connection is not a sales opportunity. It's the beginning of a conversation."

The third stage: when it seems real

Today we're in a third stage, and it's harder to detect because sometimes it sounds genuine.

AI can analyze your profile, your posts, your trajectory and your language, and generate in seconds a message that seems written for you. The volume didn't disappear. It disguised itself as attention.

The fundamental error

"Personalizing on the person no longer works. Personalize on the trigger."

The problem isn't using AI to reach out. It's using it to simulate a relationship that doesn't exist as a shortcut to something you want to obtain. And that can be felt.

In overly elaborate, flattering or strategic messages. Messages that don't open conversation or build interest before asking. Real curiosity can't be automated.

Optimizing without understanding

There's also a structural problem. When every message is different, you stop understanding what works. You can't measure what generated the response. You're optimizing… blindly.

Individual AI personalization tends to perform worse than messages designed at the segment level. Not because they're worse written, but because they optimize on the wrong layer.

It's not about writing better to more people. It's about writing to the right people, at the right time, with something that actually makes sense for them.

The shift in approach

Personalization that works doesn't start with the person. It starts with the situation. Not what they posted or how they describe themselves, but what they're probably going through now. What change they're experiencing. What problem they might have today.

When segmentation is precise, personalization becomes almost optional. A simple message works better because the person recognizes themselves in the problem or genuine interest is awakened. It's not copywriting. It's fit.

On LinkedIn: first open. Then advance.

A clear, well-built profile with relevant content isn't aesthetics. It's infrastructure. It defines whether someone understands in seconds who you are, what you do, and why it should matter to them.

Knocking on the door with a message when connecting isn't a detail. It's intention.

An accepted connection isn't a sales opportunity. It's an opportunity for conversation. A connection opens the door. Relevance defines whether it stays open.

Article originally published by Carla Fresch Pons on LinkedIn.