Personalisation that doesn't feel creepy

Specificity that implies surveillance. A user who browsed a product page once and then opened their email to "We noticed you were looking at the navy jacket" gets an instant discomfort reaction — not because the data is wrong, but because the phrasing makes the observation explicit. The same data point delivered as a product recommendation in a digest feels fine. One reads as being watched. The other reads as being served. Identical data. Completely different experience.

The rule most teams miss: the creepy line isn't about what you know, it's about how you signal what you know. A marketing team showing off their data infrastructure in the copy is the fastest way to trip the reaction. A team using the same data silently to improve relevance is usually fine.

This is a voice and copy discipline more than a data one. You can have every user's complete behavioural history and never trigger the reaction, provided your copy treats the data as invisible context rather than a headline.

Relevance without reference. "Here are jackets we think you'll like" works; "Based on the jacket you viewed on Tuesday" doesn't. The recommendation is personalised, but the copy doesn't narrate the personalisation. Users register the relevance without being reminded they're being tracked. Silent relevance is the default mode for nearly every message you send.

Recent product activity as context, not subject. A cart-abandonment email referencing "your cart" feels fine because the cart is a user-initiated state the user created. A browse-abandonment email referencing "that item you looked at" feels less fine because browsing isn't an action the user chose to take. Commitment the user made — fair game. Passive behaviour the user didn't commit to — danger zone. Is browse-abandonment creepy? Often yes. Cart-abandonment often no. Test both; cart-abandonment conversion is typically higher anyway, so you're rarely sacrificing performance to err on the right side.

Aggregate patterns framed as trends. "Customers who bought X often also buy Y" works as social proof. "We noticed you bought X" reads as surveillance even though the underlying data point is identical. Framing personalisation as belonging to a group instead of an individual feels safer because it is — the individual isn't singled out by name.

Transparent, user-controlled settings. A preference centre that lets the user say "show me more running gear" turns personalisation into collaboration. The user knows why they're seeing what they're seeing and can change it. Counterintuitively, programs with strong preference centres often have higher tolerance for data-driven content — the user has signed on to the system and knows where the dials are. The Liquid reference covers the syntax for implementing preference-aware personalisation in Braze specifically.

Cross-device narration. "We saw you looked at this on your phone yesterday and thought you might want it on your laptop too" — even if true, this reads as proof the system tracks users across devices. Users mostly accept that tracking happens; very few accept being told about it by name. Should you reference products the user looked at across devices? Only silently. Show the recommendation. Don't narrate where it came from. Use the data; don't announce the data.

Re-surfacing abandoned data. A user deletes an item from their cart. Un-favourites a product. Navigates away deliberately. An email the next day referencing that item reads as ignoring the signal the user sent. The rule: honour negative actions at least as much as positive ones. If the user said "no" through behaviour, the lifecycle program has to hear it. Otherwise the message reads as "we weren't listening", which is worse than not personalising at all.

Implied life events. "We noticed you've been browsing wedding dresses — here are more" assumes a life context the user hasn't volunteered. If they were shopping for someone else, or the engagement ended, or the browse was idle curiosity, the email becomes memorable for the wrong reasons entirely. Any personalisation based on inferred life events needs explicit user consent and an escape hatch. Most programs should just not do it.

The teams that ship the best personalisation often use less data, not more. A clean, explicit user-preference dataset produces better targeting than a sprawling behavioural profile — because the signals are reliable, the user can be shown them without embarrassment, and the compliance surface stays manageable.

Most lifecycle programs over-collect and under-use their data. The CRM Data Model skill covers the discipline of picking the smallest dataset that supports your personalisation strategy. You can always collect more. Removing data already in the system is much harder, and the compliance surface of unused data keeps expanding as regulations tighten. Should you collect less? Usually, yes. Clean explicit preferences outperform sprawling behavioural datasets on almost every dimension that matters.

A useful internal test: for every piece of personalisation in your program, can you show the user exactly which data produced the message, and would they be comfortable with that view? If the answer to either is no, reconsider. The test catches most of the decisions that create creepy moments, usually before they ship.

Sensitive contexts. Grief, medical situations, legal processes, financial distress — all contexts where personalisation reads as insensitive even when technically appropriate. A user who stops placing orders because they're going through something serious does not need a reactivation email asking what happened. The right behaviour is to pause marketing sends entirely — possible via quieted segments or explicit user flags — and resume only when the user signals readiness. The discipline is knowing when to fall back to unpersonalised content, or no content at all.

New users. The first few messages to a newly-acquired subscriber should lean on explicit preferences (what they opted into) rather than inferred behaviour (what they've done in the product). Behavioural personalisation assumes you've earned the right to reference the user's actions. The first few messages are where that right is still being negotiated, and referencing behaviour too early makes the negotiation go sideways.

Cross-audience sends. When a single message goes to multiple audience tiers simultaneously, personalisation that's fine for some users is awkward for others. A message referencing "your last purchase" lands well for a recent customer and strangely for a dormant one who forgot they ever bought anything. When audiences split, split the message. A slightly different subject line and a slightly different opener is usually enough to make the send work for both cohorts; a single message optimised for neither ends up awkward for both.