Google: Search Mobile OSRP

UI refresh and color system design for color extraction in Organized Search Page Result (OSRP) mastheads. Color extracts from images relating to search results and snaps to a systematized palette with predetermined values for various elements of the UI. Our team’s goal was to modernize the product and create a system that is visually appealing while maintaining product performance and accessibility.

Role
Visual Designer

Tools
Sketch, Color Contrast Analyzer

Platform
Google Search

Client
Google

Year
2019

Team
3 Designers, 2 Researcher, 4 Engineers

Overview

A dynamic color-extraction system for Google Search’s OSRP mastheads, enabling UI palette adaptation to third-party imagery. Utilizing a radial-snapping algorithm, we extract dominant hues from masthead assets and map them to a disciplined token scale (C1–C8, CX), ensuring consistent application across dividers, backgrounds, tabs, and CTAs.

The system supports over 20 distinct hue inputs, and upholds 100% WCAG AA compliance on follow buttons, active tabs, and links. Validated through hill-climbing A/B experiments over 10 million daily queries—each variant held within a ±1% click-through regression guardrail—the chosen CX-row configuration achieved a +0.2% engagement lift.


The Challenge

  • Monolithic Brand Blue: Years of performance tuning had locked every tab, button, and divider into a single dark-blue, ignoring context.

  • Visual Dissonance: Rich third-party imagery felt disconnected from the static UI, eroding perceived polish.

  • High Stakes: Any tweak risked measurable click-through regressions beyond our ±1 % threshold.

  • Accessibility Mandate: Must meet WCAG AA/AAA for Follow buttons, active tabs, and tappable text across 20 extracted hues.


Accessibilty Audit

Audited 21 extracted hues × 5 UI elements in an automated contrast matrix—identifying ~40% of combinations failing AA.

Key Insights:

  • Accessibility Gaps: One-third of raw hues failed AA on interactive tabs, as surfaced by our audit and a post-launch bug report.

Solution:

  • CX Row: A custom CX row of eight optimized hues met AA/AAA, felt on-brand warm, and became our standard palette.

Hillclimbing for Perfomance

Designed and ran four hill-climbing experiments over two weeks, testing contrast boosts, gradient intensity, and tab styles across 10 million daily queries—measuring click-through deltas against our ±1 % guardrail.

Key Insights:

  • Contrast vs. Clarity: A +20 % contrast arm yielded a +0.3 % engagement lift (well within tolerance) but introduced a “muddied” look flagged by observers.


Goals & Success Metrics

  1. Harmonize UI with any masthead imagery

  2. Maintain click-through within ±1 % (target: 0 %)

  3. Achieve 100 % WCAG AA on interactive elements

  4. Ship a reusable, documented token library


Final Specs

Building on these findings, I documented:

  1. Radial-Snapping Documentation

    • Fetch masthead asset → extract dominant hue → snap to nearest token (C1–C8 or CX) → apply via CSS/Android/iOS variable.

  2. Token Mapping & Usage

    • C3: Hairlines & dividers

    • C5: Background gradients

    • CX: Interactive elements (tabs, Follow CTA)

    • C8: Tappable text links


Impact

Within three weeks of rollout:

  • 80 % Adoption of dynamic palettes across global OSRP mastheads

  • 100 % WCAG AA coverage on key interactive elements (up from 68 %)

  • +0.2 % Click-Through Lift on the chosen CX-row variant

  • Qualitative Feedback:

    “This dynamic palette feels remarkably on-brand yet fresh,” — VisD lead.

System Overview

Audited 20 extracted hues × 5 UI elements in an automated contrast matrix—identifying ~40% of combinations failing AA.

Key Insights:

  • Accessibility Gaps: One-third of raw hues failed AA on interactive tabs, as surfaced by our audit and a post-launch bug report.

  • CX Row: A custom CX row of eight optimized hues met AA/AAA, felt on-brand warm, and became our standard palette.