Keep what is used often, costly to miss or next in the task on screen; move the rest one labelled step away, and test it with five real tasks. Progressive disclosure means showing people the controls and information their current task needs, and keeping the rest one deliberate step away.
Progressive disclosure means showing people the controls and information their current task needs, and keeping the rest one deliberate step away. On a dense B2B screen the technique is easy. The decision is hard: what stays visible, what moves one step away, and what that move costs. We make it with three questions about every control. How often is it needed? What does it cost to miss? Does it come next in the task? Anything in frequent use, costly to miss or next in sequence stays on the screen. Everything else can move behind a disclosure, and each move gets checked against real tasks before it ships.
This page is for the product manager who has to say no when every control has a sponsor, and for the designer working beside them. It covers where the idea comes from, a grid for placing each control, the disclosure patterns and what each one costs, the ways disclosure fails, and a five-task test for checking the result.
What progressive disclosure is, and where the idea comes from
Progressive disclosure is the practice of sequencing an interface's function: what a person meets first, what they reach when they ask for it, and what waits until a later step. Nothing is removed. The order changes.
A clear early statement comes from the team behind the Xerox Star workstation. In Designing the Star User Interface (BYTE, April 1982), David Canfield Smith, Charles Irby, Ralph Kimball, Bill Verplank and Eric Harslem explain that Star had a couple of hundred object properties, yet each property sheet showed only those relevant to the selected object. They called it an example of progressive disclosure, "hiding complexity until it is needed". The same article quotes Alan Kay's maxim that simple things should be simple and complex things should be possible, and admits the trade it involved: common tasks were sometimes made simpler at the cost of uncommon ones.
Two years later, John M. Carroll and Caroline Carrithers at IBM tested a stronger version. In Training wheels in a user interface (Communications of the ACM, August 1984), a training version of a commercial word processor made typical, troublesome error states unreachable for new users. Learners on it learned substantially faster and did better on a comprehension test, while a control group spent almost a quarter of its time recovering from the errors the training version had blocked.
Two lessons carry over to B2B software. Disclosure is a decision about the order in which function meets the user, not about how much function exists. And the strongest evidence for hiding comes from people who are learning. If your users spend their working day in the product, they are past the learning stage, and that changes the bet.
Simple is not fewer things
A request we hear from product teams is a simpler screen, with nothing on it allowed to go. Both halves can be right. On a dense screen, simple means the right things in the right order. Every control you hide is a bet that the person will not need it now. Good disclosure makes those bets on purpose, from task evidence, with a way back for the expert.
Users are divided about removal. Joanna McGrenere and Gale Moore studied 53 people who used Microsoft Word 97 for Are We All In the Same "Bloat"? (Graphics Interface 2000). Only 24.5% agreed that they wanted just the functions they used. 45.3% preferred unused functions tucked away, and 50.9% said that continually discovering new functions mattered to them. Cutting function pleases a minority. Moving it out of the way, with a route back, pleases more.
We met the same tension in our redesign of Dealpath's listing card. A listing carried units, square footage, year built, price guidance, building class, occupancy and investment profile, because acquisitions teams need all of it to judge an opportunity. Removing fields would have made the screen cleaner and the product less useful. We gave the card three zones and a reading order instead, and the same information stayed available.
The decision rule: frequency, consequence and sequence
Ask three questions of every action and every piece of information on the screen, and answer them with evidence rather than opinion.
- Frequency. How often is it used, and by which role? Use product analytics event counts where they exist, and watch people work where they do not. For a new product, observe the task as it is done now, in spreadsheets, email or an older tool.
- Consequence. What happens if someone misses it, or cannot find it when they need it? Missing that a payment is on hold is costly. Missing a column-reorder control is not. Support tickets, incident reviews and error logs show where the cost falls.
- Sequence. Does it come next in the task the person is doing now? A control that is the obvious next step earns a place in that state, even if it is rare across the product as a whole.
Frequency and consequence place a control in one of three tiers: always visible, one step away, or tucked away. Sequence promotes it for the moment it matters.
| Costly to miss | Cheap to miss | |
|---|---|---|
| High frequency | Always visible, in a fixed place | Visible but quiet: an icon button, a keyboard shortcut, compact spacing |
| Low frequency | Its state always visible; the action one labelled step away, promoted when it comes next | Tucked away: an overflow menu, settings or a saved view |
Four rules settle most arguments the grid leaves open.
- Information and actions are placed separately. If missing a piece of information is costly (a status, a scope, an applied filter, a warning), the information stays visible even when the action that changes it is tucked away. IBM's Carbon disclosures guidance says the same: do not hide information a person may need to complete a task.
- Sequence beats frequency for the moment. "Submit for approval" may be used once per record, but once the record is complete it is the next step, so it belongs where the eye already is.
- Role decides frequency. The same control can be daily for an administrator and yearly for a reviewer. Count frequency per role, not as one average across everyone.
- Rare destructive actions are safer one step away. Distance is a guard. Carbon's common actions guidance scales the guard with impact: no warning for a trivial deletion, a confirmation for one that cannot easily be undone, and typing the resource's name for a high-impact one.
Mature design systems encode the same logic. SAP Fiori's toolbar guidance sorts actions from most to least frequently used, so the most important are the last to move into the overflow menu. It lets teams mark an action "Never overflow" or "Low", and it leaves the call about which actions are important to the app team. Carbon's data table guidance allows up to five actions in a table toolbar and puts the rest in an overflow menu or similar control.
The same three questions decide what goes where across a whole product, not just one screen. That wider job is information architecture and workflow design.
Disclosure patterns and what each one costs
Every disclosure pattern trades space on the screen for effort, memory or visibility somewhere else. Choose the pattern by the cost you can afford.
| Pattern | Use it when | What it costs |
|---|---|---|
| Overflow menu | Rare actions on an object or a toolbar. Atlassian's dropdown menu guidance suggests 5 to 15 items, the most selected first, grouped under titles | Its contents are invisible until opened, and an icon-only trigger says nothing about them. It also attracts every disputed control |
| Expandable section or row | Supplementary detail read now and then. Carbon uses expandable table rows for supplementary information | Collapsed content is not scanned or compared across rows. When the expanded area feels cramped, Carbon points to a side panel or a dedicated page |
| Inspector panel | Detail of the selected item while the list stays in view | It takes width from the list, and what sits in it cannot be compared across items |
| Staged form | Long input where later questions depend on earlier answers | It hides the length of the job and what is coming. Show the steps ahead, and let people go back without losing work |
| Hover and tooltip | A definition, or the full text of a truncated label | It disappears when the pointer moves, and a touchscreen has no hover. Carbon's tooltip guidance rules out information needed to complete a task, and any interactive element |
| Saved views | Recurring sets of filters, columns and sorting for a person or a team | Someone has to maintain them, and a shared view can change under the people using it. Show what a view applies, so nobody works on filtered data without knowing |
| Role-based defaults | Roles with genuinely different jobs on the same object | People who hold two roles, and support teams who see a different screen from the customer. Keep a route to the full set |
Density is the lever teams forget. Salesforce lets an organisation set a default display density, comfy or compact, and lets each user set their own from the profile menu. Tighter spacing can keep a control visible that would otherwise have to move. For how density, saved views and bulk actions work inside a grid, see our guide to data table design for B2B SaaS.
Whatever the pattern, the trigger carries the weight. GitHub's Primer asks teams to use disclosure sparingly and to pair disclosure icons with descriptive text. Carbon opens one disclosure at a time and avoids nesting one inside another, and Atlassian limits nested menus to two layers where possible. Build the trigger as the W3C's disclosure pattern describes: a button that reports whether its content is expanded, and that Enter and Space both toggle.
How progressive disclosure fails
Disclosure on a dense screen fails in four common ways.
Hidden actions nobody finds. The symptom is a feature request for something that already exists. Hiding by rules the user cannot see makes it worse. In a field study by Joanna McGrenere, Ronald Baecker and Kellogg Booth, An evaluation of a multiple interface design solution for bloated software (CHI 2002), 7 of 20 participants had to be told that Word 2000's short menus were adapting to their own usage. Hide by stable rules, label the trigger, and follow SAP's instruction to place commands in the same location throughout the app.
The "Advanced" drawer that becomes a junk drawer. Every control nobody could agree on lands there, because moving it ends the meeting. A drawer named for its contents, grouped and capped, stays useful. Atlassian's guidance is to test and refine over time and re-evaluate whether every menu item is needed. Give each item an owner and a usage count. If nobody opens an item for a full release cycle, ask whether it should exist at all.
The expert slowed by clicks. The training-wheels evidence was about learners. A disclosure that spares a new user a moment of confusion costs a daily user one more click on every record, every day. Build the way back with keyboard shortcuts, saved views, a density setting the user controls, and a quick route to the full set. McGrenere, Baecker and Booth gave users a personal interface they built themselves and a full one, switched with a single button click; 13 of their 20 participants preferred it to either form of Word 2000. Serving novices and experts in one interface is a subject of its own, covered in designing one interface for novice and expert users.
Hidden state. A filter, scope or hold that is applied but collapsed out of sight makes wrong data look right. Anything that changes what a person is looking at stays visible, even when the controls that change it are tucked away.
Test the decision with five real tasks
A placement is a hypothesis until people have worked through it. Before anything ships, write each decision down: the control, the role, the frequency and consequence evidence, the tier, and what would make you move it. Then test it with five tasks, one of each kind.
- The most frequent task for the main role. Pass: done without opening anything.
- The most consequential task, where a mistake costs money, data or trust. Pass: everything needed to judge it is visible at the moment of decision.
- A rare task that cannot wait, such as correcting a record that went out with an error. Pass: the person finds the control without help.
- An expert task at speed, done by a daily user. Pass: no more steps than before, or a faster route through a shortcut or a saved view.
- A recovery task. Something on screen looks wrong, and the person has to find out why: an applied filter, a scope, a hidden column. Pass: the cause is found from what is on the screen.
Run it with people from each role, both new and experienced, on realistic data. Word each task as a goal, never as the name of the control. Record where people look first, whether they open the right disclosure, how many steps they take and what they get wrong.
Then move things. Anything missed in tasks 1, 2 or 5 moves up a tier. Anything nobody opened is a candidate to move down, or to leave the product. After release, count how often each disclosure is opened and what is chosen inside it, and review the placements on the same cycle as the drawer.
Related decisions
Progressive disclosure is one part of making complex software easier to work in. For the wider principles, read our guide to enterprise UX design for complex workflow software. If your team is weighing what a crowded screen should show, this is the work of our interaction and UI design practice.
About the authors
Yash Gidwani
Product Designer
More from YashYash Gidwani is a Product Designer at Tcules who writes about getting people into a SaaS product and keeping them there. Yash's guides cover SaaS onboarding strategies, onboarding and engagement software, and redesigning a live product without upsetting the people who use it. Other articles go into specific workflows: lending and loan origination apps, insurance quoting and claims, staffing and recruitment software, and a search-first create flow that stops duplicate records. Yash has also co-written the SaaS UX audit checklist and a 20-question checklist for legacy modernisation readiness.

Rushali Bhatt
Product Designer
More from RushaliRushali Bhatt is a Product Designer at Tcules. Rushali's articles start from the people inside complex software: clinicians and patients in healthcare products, users who cannot take part in research sessions, and the reviewers who check an AI feature's output before anyone acts on it. Topics include usability studies for enterprise software, human-in-the-loop patterns, a readiness checklist for AI product UX, what to fix first when adding AI to a live SaaS product, and what a heuristic evaluation should deliver. Useful reading for product teams whose software people rely on to make decisions.