‹ back to portfolio

Windows interaction design

Windows Privacy Settings

Know what has access.

Making privacy permissions easier to discover and tune, with clear categories and app-access counts at a glance.

Shipped redesign · 4M+ monthly active devices
The redesigned Windows Privacy and security page, with grouped permissions and app-access counts
My role
Software Engineer · Interaction Design
Collaboration
Designers and PMs
Scope
Windows privacy & security · L1 refresh

The problem, in one minute

People could find the permissions page, but still had to open individual settings to understand which apps had access. CSAT verbatims helped explain a behavior visible in telemetry: arriving, then leaving without changing anything.

I contributed interaction designs informed by those signals and implemented the architectural and rendering changes. The focus was clearer hierarchy, visible access information, and trustworthy loading behavior — not adding another set of controls.

+10%
feature discoverability
−1.57s
perceived load time
6 groups
replacing one flat list

These outcomes describe the shipped refresh, combining design and engineering changes; they are not results from an isolated comparison of the diagrams below.

01 / Information architecture

Group by concern, not a flat inventory

The choice: organize permissions around what they protect. A person worried about their microphone should not need to understand the underlying API categories. Location, camera, and microphone become a prominent core group; the remaining controls gain named destinations.

Before: one list to scan

  • Location
  • Camera
  • Microphone
  • Voice activation
  • Notifications
  • Account info
  • More permissions…

After: a hierarchy to navigate

  • Core privacy settingsLocation · Camera · Microphone
  • Security
  • Your info
  • Communication
  • File access
  • System
Schematic comparison of the described structures, not original iteration files. The goal is faster orientation, not fewer controls.
The shipped Privacy and security page with Core privacy settings above a separate Security group
  1. Core controls first. High-stakes permissions get a clear starting point.
  2. Security stays distinct. Related protections remain nearby without becoming one undifferentiated category.
Shipped interface, with annotations added for this case study.

Trade-off: grouping adds headings and demands meaningful labels. Reusing the same row component keeps scanning consistent, so visitors learn one interaction pattern rather than a different layout for each category.

02 / Access visibility

Put the answer in the scan path

The choice: show an access count before someone opens a permission. The screenshot makes the change concrete: each core row now answers “how many apps?” A count summarizes access; it does not mean those apps are using the sensor right now.

Core privacy rows showing 12 apps for Location, 8 apps for Camera, and 6 apps for Microphone
Shipped detail: permission names and access counts can be scanned together, without opening three separate pages.
  • Loading

    Location

    Reserve space; do not imply zero.

  • Access off

    LocationAccess off

    Permission disabled across the system.

  • Zero

    Location0 apps

    Enabled, but no apps granted access.

  • One

    Location1 app

    Singular phrasing matches the count.

  • Many

    Location12 apps

    Summarize before showing app details.

Illustrations of the states described in this case study, not captured product screens. The sample values explain behavior rather than document separate iterations.

Trade-off: a concise summary can hide important distinctions. Loading, disabled access, and an empty app list must not look interchangeable. Stable placeholders and explicit wording make those differences visible while retaining a route to the underlying details.

03 / Layout under pressure

Let the content set the height

The change: move away from a header sized for a comfortable English string. Spanish and Japanese checks exposed different text lengths and shapes. The component needed to absorb another line without clipping the message or displacing its neighboring controls.

Fixed-height assumption

12 applications are allowed to access your microphone

Expanded text is clipped.

Content-driven height

12 applications are allowed to access your microphone

Text wraps; the control stays aligned.

Illustrative stress test using expanded English copy, not a translated product screenshot or historical prototype. The controls here are explanatory, not interactive.

Trade-off: flexible rows take more vertical space, but preserve meaning across languages. Plural forms and word order belong in the component design, not in a late patch that shortens or hides the message.

What I would carry forward

The rendering work supported the same interaction goal: present a stable, understandable page promptly. Contributing to the design and implementing it kept performance constraints in the conversation while the component could still change.

  • Design the non-happy paths early. Empty, disabled, loading, and expanded-text states reveal whether the component communicates reliably, not just whether the ideal screen looks finished.
  • Pair behavioral signals with people's words. Telemetry can flag where visitors stop; verbatims help explain why. Together they make the next design question more specific.

Screens show the public-facing Windows experience. Internal planning artifacts and unreleased surfaces are not included. Diagrams explain the documented decisions; they are not original design-history evidence.