Project Overview

When Data Becomes an Obstacle

Yahoo!'s Demand-Side Platform gives advertisers the power to plan, purchase, and optimize ad campaigns, but its analytics dashboard had become a maze of unfiltered data, slow load times, and frustrating limitations that left users disengaged and inefficient. This was an 11-month solo design project - I owned every phase, from the first user interview to final design handoff.

1
Solo UX Designer
11mo
Project duration
14
User interviews
5
Design thinking phases
My Role: Senior UX Designer · Solo
User research & live observation
Dashboard & data visualization design
Information architecture
End-to-end prototyping
Usability testing & iteration
Stakeholder alignment & handoff
Data Visualization Dashboard Design User Research Information Architecture Visual Hierarchy PEARL Framework

Design Process

Design Thinking: End to End

Phase 01
Empathize
14 user interviews and live observation sessions with Yahoo! DSP account strategists uncovered the real problem: a senior strategist spent 5+ minutes navigating menus before saying "I spend more time finding campaigns than actually analyzing" - the moment that reframed the entire project.
Phase 02
Define
Four failures defined the scope: no filtering system across hundreds of campaigns, no charts or trend signals on raw data, 9+ second average load times, and no customization for different advertiser workflows. Each one amplified the others.
Phase 03
Ideate
Three design principles governed every decision - Clarity & Simplicity, Visual Hierarchy, and Timely Feedback - structured around progressive disclosure: surface the most critical insight immediately, with deeper data available on demand rather than all at once.
Phase 04
Prototype
Three rounds of prototyping - paper sketches, mid-fidelity wireframes, full Figma - tested with 3–5 users at each stage. The defining decision: a chart-first layout replaced raw data tables as the primary view, with detailed breakdowns accessible in a slide-out panel.
Phase 05
Test
A 6-week post-launch measurement cycle using Google Analytics and Mixpanel: –33% data retrieval time, +25% engagement, +34% satisfaction, and a user rating jump from 3.5 to 4.7. The dashboard architecture was later adopted by two other Yahoo! products: SSP and Flurry.

Phase 1: Empathize

A Moment That Changed Everything

The turning point came during a live observation session. I watched a senior account strategist spend over five minutes navigating menus and filter combinations, before finally muttering something that stopped me cold:

"I spend more time trying to find campaigns and navigating them than actually analyzing."
, Senior Account Strategist, Yahoo! DSP · Live observation session

That one sentence reframed the entire problem. This wasn't about missing features. It was about a dashboard that got in the way of insight.

I followed up with structured research across three methods:

User Interviews (n=14) Live Observation Sessions Usability Testing Stakeholder Workshops Analytics Review (Mixpanel)

Empathy Map: What users were feeling:

Persona: Who I was designing for:

Phase 2: Define

Naming the Problem with Precision

The research synthesis revealed a dashboard failing in four fundamental ways, each one amplifying the others in a spiral of frustration:

  • No filtering system: Users were forced to scroll through hundreds of campaigns with no way to isolate what mattered.
  • No data summaries or visualizations: Raw numbers with no context, no trends, no comparative signals. Users couldn't tell at a glance whether a campaign was performing or failing.
  • Slow performance: Average load time of 9+ seconds created constant friction and broke concentration.
  • Limited customization: One-size-fits-all views ignored that different advertisers prioritize different metrics and workflows.
How might we redesign Yahoo!'s Ads Data Dashboard to give account strategists the clarity, speed, and control they need to move from data to insight, in seconds, not minutes?

Problem Statement (validated with stakeholders):

Yahoo!'s Ads Data Dashboard lacked filtering, visualization, performance, and customization, causing user frustration, reduced engagement, and systematic inefficiency in campaign optimization workflows.
Phase 3: Ideate

Designing for Clarity, Speed, and Control

With a clear problem definition and a validated HMW statement, I moved into ideation. I sketched and wireframed layouts focused on progressive disclosure, surfacing the most important information immediately, with deeper data available on demand.

Three design principles governed every decision:

Principle 1: Clarity & Simplicity

Information must be digestible at a glance. Every element on screen must earn its place. Ruthless prioritization. If it doesn't help a user make a decision, it doesn't appear by default.

Principle 2: Visual Hierarchy

Size, color, and layout were used strategically to guide users to high-value insights first. Critical performance indicators were elevated; secondary data was contextually accessible but not dominant.

Principle 3: Timely Feedback

Every user action, filtering, sorting, date range selection, triggers immediate, visible feedback. Users always know the system heard them and is responding. Uncertainty kills trust.

UX Laws Applied:

  • Fitts's Law: Key action elements (filter button, date picker, export) were sized and positioned for minimum movement and maximum speed.
  • Jakob's Law: The redesigned dashboard aligned with familiar data visualization patterns (chart types, filter behavior, drill-down logic) so advertisers didn't need to relearn.
Phase 4: Prototype

From Sketches to High-Fidelity Designs

I moved through three rounds of prototyping, paper sketches, mid-fidelity wireframes, then full Figma prototypes, testing at each stage with 3–5 users before progressing. This kept iteration cheap and insight high.

Prototype iterations:

Early wireframe: navigation & filter exploration

Mid-fidelity: chart hierarchy & data modules

Key design decisions in the final prototype:

  • Smart contextual filters: Filter logic adapts to the data type being viewed, surfacing relevant options automatically.
  • Chart-first layout: Replaced raw data tables as the primary view with interactive charts; raw data is available in a slide-out panel.
  • Customizable dashboard modules: Users can choose which data blocks appear on their primary view, saved per user profile.
  • Progressive data loading: Critical summary metrics load first; detailed breakdowns load progressively to eliminate the perception of slowness.

Final design: live dashboard interaction:

Phase 5: Test & Impact

Numbers Don't Lie

Post-launch, we ran a 6-week measurement cycle using Google Analytics, Mixpanel funnels, and structured survey instruments. The results were immediate and sustained.

+25%Increase in user engagement
–33%Reduction in data retrieval time
+34%Satisfaction score improvement
3.5→4.7User rating (out of 5)

Beyond the numbers. The redesigned dashboard became an internal benchmark. Two other Yahoo! analytics products (SSP and Flurry) adopted its navigation architecture and visualization patterns, multiplying the project's impact beyond its original scope.

"The dashboard actually shows me what I need now. I can see a campaign's health in seconds. I couldn't do that before."
, Account Strategist, post-launch feedback survey
Design Leadership

Calls I Made on This Project

Decision I owned
Changed the entry point from time-picker to KPI cards

The original dashboard opened with a date selector - time-first logic, inherited from a reporting mindset. Research showed users arrived with a performance question in mind: "Is this campaign working?" That's an outcome question, not a time question. I redesigned the entry point around four KPI cards so the first thing a user sees is the answer to their actual question. This was a structural change that required re-scoping 3 weeks of design work. It was the right call.

What I pushed back on
The "data wall" approach in the original spec

The original product spec called for surfacing maximum data above the fold - the theory being that power users want everything visible at once. The epiphany session I ran (14 interviews, 5:48 average task time) showed the opposite: users were overwhelmed, not empowered. High time-on-screen was friction, not engagement. I pushed back on data density as a design goal and introduced a three-tier visual hierarchy: critical KPIs → trends → raw data, with each tier de-emphasized progressively.

What got cut and why
The custom dashboard builder

A drag-and-drop dashboard customization layer was prototyped mid-project and was technically feasible. I cut it. The whole thesis of the redesign was that users shouldn't have to configure their way to insight - the right data should be there by default. Shipping a dashboard builder would have contradicted that argument and shifted the work of information architecture onto users who already told us they were overwhelmed. Opinionated defaults won. We'd revisit configurability only if post-launch data showed users actively needed it.

Reflection

The Core Lesson

Good design doesn't add features. It removes friction.

This project validated a principle I return to constantly: the most powerful design decisions are subtractive, not additive. By deeply listening to users, not stakeholders, not assumptions, and letting empathy drive every design choice, I transformed a bloated, underused tool into a high-performance product that people actually wanted to use.

The PEARL framework (Problem, Epiphany, Action, Result, Learning) kept the story focused and human across 11 months of iteration. Design thinking isn't a flowchart. It's a commitment to staying curious, staying humble, and staying user-first at every step.

Interested in the full design documentation? Connect on LinkedIn/Nancy-UX for access to wireframe progression, testing transcripts, and analytics breakdowns.

User Feedback

What Users Said

Direct feedback collected through post-launch surveys and follow-up interviews.

Account Strategist

"I used to open five different reports just to understand one campaign. Now everything I need is in a single view - it's like someone finally listened."

Post-launch feedback survey · Yahoo! Ads
Media Planner

"The charts used to feel decorative. Now they actually tell me where to act. The trend lines alone have changed how I structure my morning reviews."

Usability follow-up interview · 6 weeks post-launch
Account Director

"We've been asking for this for two years. I can finally stop second-guessing the data and focus on the decisions that actually move the needle."

Quarterly product review · Yahoo! Ads leadership team
Previous
Case Study
eBay Magical Listings
Next
Case Study
Yahoo! Ads Tables