Progressive disclosure: show only what is needed
Progressive disclosure shows the essentials first and moves advanced options a click away. What it is, when it works and how to avoid hiding what matters.
What is progressive disclosure?
It is a way of organizing information in layers. The first layer holds what most people need to complete a task. The following layers hold what appears in specific cases: fine-grained settings, exceptions, history, or features used once a month.
Jakob Nielsen defines it in his 2006 article for Nielsen Norman Group with a short sentence: the technique defers advanced or rarely used features to a secondary screen. Interaction Design Foundation describes it in similar terms: it postpones advanced features and information to secondary interface components that the user opens when needed.
An example everyone knows is the print dialog. The first screen shows the number of copies, which pages to print and which printer to use. An advanced options button opens a secondary dialog with rarely used settings, such as scaling or reverse page order. Nielsen uses it as the classic example of the pattern, and notes that some print dialogs have become bloated and should move more detail into secondary dialogs.
Why does it matter to a designer?
Because every visible element competes for attention. When a screen shows twenty controls, the user has to decide which ones to ignore before they can act. Progressive disclosure lowers that decision cost.
It also reduces errors. An option almost nobody needs, placed next to the main action, is a chance to change something by accident. Moving it to a secondary level makes the frequent action safer.
And it makes the product easier to learn. Someone starting out does not need to understand the whole system to finish their first task. They can learn the advanced features when those features start to make sense to them.
How do you apply it without hiding what matters?
The most common mistake is deciding what to hide by intuition. Nielsen argues that the split between the initial and secondary levels should be based on data: for a new design, task analysis and field studies; for an existing system, how often each feature is used, measured with analytics or by instrumenting the code. He also warns that on a website, page hits can be accidental, so the numbers should be supplemented with observational usability testing. Interaction Design Foundation proposes something similar, using card sorting and task analysis to decide what stays in the main view.
In practice, follow these steps:
- List every function on the screen or flow, with its estimated frequency of use.
- Separate what most people need for each task. That goes on the first level.
- Put the rest on a secondary level, with a label that says what it contains. Nielsen asks that the control for moving forward make clear what is on the next level (what he calls strong information scent). “Advanced colour settings” says more than “More”.
- Keep the path visible. A user looking for something must be able to find how to reach the next layer without guessing.
- Limit the levels. Nielsen states that designs with more than two levels typically have low usability, because people get lost moving between them; if you need three or more, simplify or group the advanced features into logical chunks. Interaction Design Foundation says a single secondary screen is typically enough.
- Test with new people and with experts. A secondary level nobody finds is of no use.
When should you NOT use it?
It is not the answer to everything. There are cases where hiding makes the experience worse:
- Decisive information. The price, cancellation terms, requirements for a procedure or deadlines should not sit behind a click. In his 2026 text, Nielsen counts hiding warnings and revealing costs late among the failures, and asks that decision-critical information appear before commitment.
- When the expert is the main user. In professional tools, a shortcut to an advanced function can be part of the daily task. Forcing three clicks to reach it costs productivity.
- When the function is used every time. Something that appears in every session is not advanced, even if the product team thinks it is.
- When there is no clear way to get there. If the access to hidden content is an unlabelled icon, the user will not find it.
Nielsen also distinguishes staged disclosure: a linear sequence of steps with a subset of options at each one, like a wizard. It works well when the task splits into distinct steps with little interaction between them, and causes problems when steps depend on each other and people have to go back and forth. It is a different technique from progressive disclosure, which leaves advanced content available on request, although they are often confused. In a payment flow, asking for data in steps is staged; a “shipping details” panel that expands is progressive.
What about AI products?
In July 2026, Nielsen extended the pattern to AI. He argues that a chat box shifts work from recognition to recall and leaves novices facing a “cliff”. For answers, he proposes giving a short verdict first and putting the reasoning and sources behind expandable controls. For agents, he proposes that the first level show the outcome and any actions needing approval, such as spending money or deleting files, and that the step-by-step log act as the advanced-settings drawer. These are one author’s proposals, not study results; use them as a starting point to test with your users. Concrete patterns are in action approval.
Example
Illustrative example: an invoicing application shows only customer, line items, amount and date on the new-invoice screen. A link labelled “Taxes, withholdings and series” opens a panel with those fields. Frequent users open it once and the panel remembers its state. New users see nothing they do not understand on their first invoice.
A test you can run tomorrow: take one screen of your product, mark each control with one of three labels (everyone uses it, some people use it, almost nobody uses it) and ask yourself whether the third group is easy enough to find.
Common mistakes
- Hiding things by the designer’s taste, without usage data or tests.
- Using vague labels such as “More” or “Other” that do not say what the secondary level contains.
- Hiding decisive information the user needs before committing.
- Creating too many levels. Three or four layers mean nobody reaches the content they were looking for.
- Giving two different paths to the same hidden content. Both Nielsen and Interaction Design Foundation recommend avoiding multiple access paths.
- Forgetting the expert user. Hiding things without a shortcut or a preference that remembers them penalizes the most frequent user.
Related in the wiki
Progressive disclosure builds on Hick’s law: fewer visible options reduce decision time. To organize information in layers, it is worth reviewing information architecture. For the steps of a flow, UI flow chunking works on the same problem from another angle. If you are designing the navigation that gives access to those levels, see also navigation patterns.