Avoiding half-baked personalisation

As marketing teams, we spend a lot of time striving for personalisation. On-site, in emails, across the products we’re building.

The underlying assumption is: customers would rather see experiences that are tailored to them than ones that aren’t. If I play tennis, I’d rather be shown tennis products than baseball products. It’s a better experience and, importantly, data shows that it tends to convert better too.

But I worry there’s an assumption that some personalisation is always better than none.

…and I’m not sure that’s true.

In fact, I think poorly executed, half-baked personalisation is often worse than not personalising at all.

The reason is that personalisation raises expectations. Once you start tailoring the experience, customers expect competence: memory, progression, judgment. If the experience doesn’t meet that bar, it can actually reflect badly on the quality of the experience as a whole.

Plus, in 2026, customers have a very high bar. They have a strong intuition for what feels thoughtful and what feels brittle. When an experience falls short of that quality bar, it’s noticeable.

A recent example

This idea crystallised for me recently when I logged into a piece of software I’ve used for several years, and know my way around it well.

After logging in, I was taken to the dashboard as usual. The dashboard had clearly been updated with new components designed to surface tutorials, guides, and other contextual information, presumably in a more personalised way.

As soon as I logged in, I received a pop-up asking me how I would like to “onboard to this software”.

Given I’m a long-term and active user of this software, the popup itself felt incongruent. Further, the several steps gathering information that I felt this company must already know about me felt even more incongruent.

I wasn’t thinking, “This isn’t personalised enough.” I was thinking, “You shouldn’t be showing me this at all.”

More than that, the experience felt brittle. The sort of brittleness that often comes with tooltips, modals, and guided flows layered on top of an existing product.

Going too wide, too fast

This kind of experience is surprisingly common.

When it comes to growth, we are naturally always looking for ways to convert more, to grow more. As a result, investing in personalisation is a no-brainer at some stage.

That said, we ought to pause and ask ourselves whether what we can effectively achieve in terms of personalisation and whether we have the resources to deliver an experience that meets our customers’ quality bar.

Personalization is typically more effort than we think, both in terms of the technology required to pull it off and then the maintenance to keep things humming smoothly, without introducing brittleness. If we build something that technically adds personalisation but detracts from the overall experience, we have to ask an uncomfortable question: have we actually made things worse?

This is a specific instance of a more general pattern: going too wide, too fast.

It’s a principle that’s often discussed in the context of early-stage startups: it’s better to build a scooter than to build one wheel of a future car. A narrow but complete product is more valuable than a shallow slice of something much larger.

The same applies to personalisation. It’s tempting to do a little bit everywhere: a popup here, a personalised module there, a few conditional flows layered on top. But shallow personalisation spreads complexity without delivering craft. It creates experiences that look sophisticated from the inside but feel undercooked to customers.

Why this happens

In my experience, half-baked personalisation is rarely the result of bad intentions. More often, it’s the product of a few structural pressures.

Firstly, time pressure. Personalisation projects often start with ambitious goals, but as complexity emerges and deadlines loom, scope gets cut. What ships is a thinner version of the original idea: something that technically works, but doesn’t quite hold up in practice.

Second, tool-driven thinking. Personalisation is frequently shaped by the capabilities of the tools we have available. A feature exists, so we look for places to use it. The focus shifts from “is this valuable?” to “how do we deploy this?”

Third, how success is measured. Metrics sometimes reward the presence of personalisation rather than the experience itself. Shipping something can feel like progress, even when there’s little time or incentive to revisit and raise the bar once it’s live.

None of this is unusual, but it does explain why shallow personalisation is so common.

A checklist for better personalisation

None of this is an argument against personalisation. But personalisation is a commitment to users, and it’s worth being deliberate about where and how that commitment is made.

When reviewing existing personalisation a few simple questions worth asking include:

• Are we asking the user something we should already know? Repeated questions signal a lack of memory, and that quickly erodes trust.
• Does this interrupt a high-intent moment? Even “helpful” personalisation can feel frustrating if it gets in the way.
• Have we explicitly decided when not to show this? Good personalisation is as much about restraint as it is about tailoring.
• Would this still be useful without personalisation? If the answer is no, you may be trying to personalise something that doesn’t deserve to exist in the first place.

These questions can help raise the quality bar.

---

Am I saying we should abandon personalisation? Absolutely not!

What I’m trying to highlight is that good personalisation, like all features of our product or store, requires investment and the sensible course. Half-baked personalisation doesn’t just fail to add value: it actively undermines trust. In many cases, a clear, stable, unpersonalised experience will outperform a brittle, shallow attempt at tailoring.

Building personalisation is generally not a one-and-done process, but something that requires commitment to the end result and therefore iteration.

Hopefully this post has given you something to think about.

Keep growing, Chris