“Recommended for you” is not personalization.
I used to believe it was.
A few years ago, I was leading a product initiative where personalization was the centerpiece of the strategy. We had the data. Behavioral segments, historical activity, demographic overlays. We built recommendation engines, tailored dashboards, and dynamic content blocks that changed based on who the user was.
On paper, it looked exactly right.
We launched expecting engagement to climb.
It didn’t.
Some users interacted with it. Most ignored it. A surprising number of them navigated around it entirely, going back to search, filters, or manual exploration. The “personalized” experience was there, but it wasn’t helping.
That was the moment I realized something uncomfortable. We hadn’t built personalization. We had built targeting.
The Challenge
What we were calling personalization was really segmentation. We grouped users into categories based on past behavior and assumed those categories defined their needs. Then we served them content we thought was relevant. The system was technically correct. But it was contextually wrong. Because people don’t behave the same way all the time.
The same user can have a completely different intent at different times. What they did yesterday is not always what they need today. But our system treated identity as static. The experience felt off. Not broken enough to complain about. Just irrelevant enough to ignore.
The Action
We stopped thinking about personalization as “who the user is” and started thinking about “what the user is trying to do right now.” That shift changed how we approached everything.
Instead of relying on segments, we started focusing on signals:
- What is the user doing in this moment
- What path are they on
- What friction are they experiencing
- What decision are they trying to make
We moved from static recommendations to dynamic guidance.
Instead of saying, “people like you usually want this,” the system began to say, “based on what you’re doing right now, this is likely the next best step.”
- We didn’t add more content.
- We reduced it.
- We didn’t try to be more accurate.
- We tried to be more relevant.
The Outcome
Engagement didn’t spike overnight, but behavior shifted. Users stopped ignoring the system and started following it. They moved faster. They made decisions with more confidence. They relied less on manual exploration and more on what was being surfaced. Not because it was personalized in the traditional sense. Because it was timely. The system felt like it understood what they needed in the moment. That’s what changed everything.
The Lesson
Most personalization today is broken. Not because the technology isn’t good enough. Because the model is wrong. We’ve been personalizing based on identity. What we should be personalizing is intent. Real personalization is not about showing users things that match their profile.
It’s about helping them in the moment they are in.
- It’s behavioral.
- It’s contextual.
- It’s real-time.
And most importantly, it reduces the need for the user to think.
“Recommended for you” is easy to build.
But it’s often just noise.
Real personalization is harder.
It requires understanding behavior as it happens, interpreting context, and guiding decisions without overwhelming the user.
It’s less about showing more.
And more about knowing what not to show.
Where This Is Going
As AI becomes more embedded in products, the gap between fake personalization and real personalization will grow.
Systems will have more data than ever. But data alone doesn’t create relevance. The products that win will not be the ones that know the most about their users. They will be the ones that respond best to what users are trying to do right now.
Final Thought
Looking back, the biggest mistake we made was assuming that if something was technically personalized, it would feel personal. It didn’t. Because personalization isn’t about the system knowing you. It’s about the system helping you.
What’s a moment in your career where something looked right on paper, but completely missed how people actually behave?
That’s usually where the real lesson is.