Personalisation has a poor reputation among the people it is aimed at, and largely for good reason. Being followed around the internet by an advertisement for a chair you already bought is not a tailored experience. Neither is an email that opens with your first name and then says nothing relevant to you.
The versions that do work tend to be quieter. They save the customer effort rather than demonstrating how much the company knows about them.
Remove steps the customer has already completed
The most valuable form of personalisation is simply not asking twice. If a returning customer has an address on file, the delivery form should already contain it. If a client has a service contract, the support page should show their contract terms without them looking it up. If someone selected a branch last visit, that branch should still be selected.
None of this feels like personalisation to the user. It feels like the company is organised. That is exactly the right impression, and it is achieved through decent session handling and a properly joined-up customer record rather than any recommendation engine.
Segment by situation, not by demographics
Age brackets and job titles are weak predictors of what someone needs from a website. What they are trying to do right now is a strong one. A first-time visitor researching options, an existing customer chasing an order, and a procurement officer comparing three suppliers all need different things from the same homepage.
Designing for two or three of these situations is achievable and worthwhile. A returning customer might see their account and recent orders in the header; a new visitor sees positioning and proof. Attempting to build twenty micro-segments produces a maintenance burden and a set of rules nobody can audit six months later.
Be careful with behavioural inference
Automated systems that guess intent from browsing behaviour are frequently wrong in ways that are awkward for the customer. Someone reading about bereavement services may be doing so for a relative. Someone comparing baby products may have had bad news. A person who looked at one expensive item once is not necessarily a luxury buyer.
Where the inference could touch anything sensitive, the safer design is to let the customer tell you. A preference they set deliberately is more accurate than a signal you inferred, and it does not carry the same risk of causing distress or appearing intrusive.
Explain and allow reversal
When content is tailored, saying so briefly builds trust rather than eroding it. “Based on your last order” next to a set of suggestions is honest and immediately understandable. A visible way to change or clear those preferences matters just as much — people share devices, circumstances change, and a profile built on last year’s behaviour can become actively unhelpful.
Under Malaysia’s Personal Data Protection Act, and for any client with customers in Europe, you also need a clear account of what data drives these decisions and a route for someone to see or delete it. Building that in from the start is much easier than retrofitting it during a customer request or an enterprise procurement review.
Keep the fallback presentable
A large share of visitors will always be anonymous: new arrivals, people in private browsing, users who declined tracking. If the default experience is a set of empty placeholders where personalised blocks should be, you have degraded the site for the audience you most need to convert.
Design the generic version first and make it good on its own terms. Treat tailoring as an improvement layered on top, never as a requirement for the page to make sense.
Judge it by customer effort
The metric that reflects whether personalisation is helping is not click-through rate. It is whether customers are completing what they came to do with less work: fewer steps to reorder from an online store, fewer support messages asking for information the site should have shown, shorter time from arrival to enquiry.
If those numbers are not moving, the tailoring is serving the company’s curiosity rather than the customer’s convenience, and it can be simplified without loss.
