Experience Etiquette: "How do you decide when personalization is useful versus surveillance?"

Dear Experience Etiquette Elana,

First, did you know your name is EEE which is basically screaming?

Second I work on lifecycle, and we’ve started leaning more heavily into personalization in our in-app messaging. Everyone talks about it constantly so… yeah. We use behavioral data, account details, usage patterns, all the bits.

It makes sense on paper. Like, I get it. Messages do better when they’re tailored. But sometimes when I read them back, I have this moment of, “If I were the customer, would this feel weird?” Like… how closely are you tracking me?

No one on the team wants to be a creep. We want to be helpful and relevant, and definitely not invasive. But that line feels fuzzy. My boss thinks the more specific we can be, the better. But I (and I’m not the only one) think we should pull back.

How do you decide when personalization is genuinely useful versus when it starts to feel like surveillance? Is there a rule of thumb?

— Not a Stalker

Dear Not a Stalker,

First of all, the fact that you're asking this question is genuinely the most important thing here. That little moment of "wait, would this feel weird to me?" is your conscience doing exactly what it's supposed to do. So good job. You're already not a stalker. Also, good job working somewhere where nobody wants to be a creep. That's, shockingly, not a given in tech right now.

Now. Let's talk about the line.

There isn't a hard rule. But there is a really good question.

When you're looking at a message and asking yourself whether it's too personal, the thing you're actually asking is: is this actually useful for this person, in the context they're in, using your platform? Does it demonstrate empathy for where they are and where they're trying to get? If not, you have to ask yourself why you're doing it at all.

That's the sanity check. Not "how much data do we have" — but "what data can we leverage to help this person do their job right now?”

The merge tag math is a red flag.

Here's a practical one: if you're looking at a message with six personalization tokens in it, you're relying on six different data points to all be accurate, up-to-date, and pulling correctly — all at once. That's a lot of faith in your data hygiene. And even if they all fire right, you have to ask whether the sum of those parts is actually adding value to the experience, or whether you've just made a very complicated sentence about someone.

More specific isn't automatically more helpful. One wrong merge tag (and it happens) erodes trust much faster than you built it. So start with intention, not volume.

Context is basically everything.

Here's a general rule of thumb that actually holds up: if a piece of personalization would feel natural in a consumer app, that doesn't mean it belongs in a B2B product. The context someone's in when they're using your platform for work is fundamentally different than when they're ordering dinner or tracking a workout. The threshold for what feels invasive at work is lower and it should be.

A good question to ask: what role does our platform play in this person's life? A fitness app that tells you how many classes you took this year? That lands because it connects to a personal goal, something someone feels good about. A work tool that tells you how many times you logged in or what features you clicked? Probably not rich enough in insight to tell them anything meaningful about the impact of those clicks or logins.

If something would only make sense in the personal context, that's a signal to really interrogate whether it belongs in the work context at all.

On managing your boss: come with an experiment, not a "no."

It sounds like what your boss is actually after is using personalization to add real value — to move people through the journey, to show empathy, to demonstrate the product gets them. That's not wrong. The disagreement is about how much, too fast.

Rather than pushing back directly, the move is to propose testing it. Start slightly broader than they want, slightly narrower than what you're worried about and set a real objective. Let the data make the case. That way you're not saying "I don't think this works," you're saying "let's find out what actually works." That's a much better position to argue from, and a much better way to protect your customers in the meantime.

The tech landscape is moving fast and everyone's feeling pressure to personalize constantly. But people are also getting better at noticing when it's unnecessary, which means the downside of overdoing it is bigger than it used to be. Start broad. Get more specific as you learn. Don't over-engineer it on the first try.

You're asking the right questions. Keep asking them.

— Experience Etiquette's Elana