# Personalization vs customization

> Customization is a change the user makes through explicit settings; personalization is a change the system makes from data about them. The distinguishing test is who chose.

_Last updated 2026-11-01. From The Personalization Guide, https://personalization.guide_

Customization and personalization both tailor an experience, and they are routinely used as
synonyms, but they sit at opposite ends of the one axis that matters: who makes the change.
Customization is tailoring the user does, through explicit settings and preferences.
Personalization is tailoring the system does on the user's behalf, inferred from data about them.
A person who picks a dark theme, pins a widget, or reorders a dashboard is customizing. A system
that reorders that same dashboard because it predicts which panels the person uses most is
personalizing.

## The distinguishing test: who chose?

Ask one question: who chose? If the user deliberately set it, it is customization. If the system
inferred it, it is personalization. The output can look identical, a dashboard in a particular
order or a homepage leading with a particular category, and still be one or the other depending on
who made the decision. That is why "who chose" beats any test based on what the change looks like.

The two also tend to feel different. Customization is predictable, because the user set it and can
set it back. Personalization can surprise, because the model, not the person, decided, and the
person may not know a decision was made at all.

## Where preference centers sit

The line is not always clean. A preference center, where someone states that they want fewer
emails, or that they care about running shoes but not formal wear, straddles it. The user
explicitly states a preference, which is customization, and the system then acts on that preference
across surfaces, which is personalization. The useful way to read these cases is to separate the
two steps: capturing a stated preference is customization; everything the system infers and does
beyond the literal statement is personalization. A system that takes "interested in running shoes"
and builds out a wider fitness profile from it has crossed from one into the other.

## Why the difference carries ethical weight

Because the two place control in different hands. Customization hands control to the person: they
set it, they see it, they can undo it. Personalization asks the person to trust the system's model
of them, a model they usually cannot see and did not build. That is not an argument against
personalization, which often helps, but it is why the two are not interchangeable in any discussion
of consent or autonomy. The questions a personalization system has to answer, such as whether the
person knows a model of them exists and can switch it off, do not even arise for customization,
where the person already holds the controls. These are the questions the
[ethics of personalization](/ethics) turns on.

## Why naming it precisely matters

Vendors blur the two because "personalized" sounds more advanced than "lets users change a
setting." But calling a preference toggle personalization hides that nothing is being inferred, and
calling an inferred change customization hides that the system, not the user, is deciding. Naming
which one a feature actually is tells you where control sits, who is accountable for the outcome,
and what the user can do about it. It is the first of several distinctions the
[comparisons](/library/comparisons) in this cluster draw, each with its own one-question test.