Launched

A Design-Led Transformation:
Make Reporting Holistic & Measurable

Design went first and shaped the product decisions. The reporting suite is now used by 100% of clients across 12,000+ campaigns. CSAT rose from 1.2 to 4.5, and replacing manual reporting saved an estimated $6.4M in operating costs.

Ad-techRevenue DrivenMajor Platform ImprovementHigh Impact Core Feature

Lead product designer · UX researcherEnd-to-end design, Strategy, Usability testing, Prototyping2 PMs · 10 engineers · 4 Looker BI engineersFeb 2024–Dec 2024

Device type donut chart
Conversion Dashboard: filters, top-line metrics, conversion funnel and details table
Conversion Dashboard: filters, top-line metrics, conversion funnel and details table
Scroll to zoom in

At a glance

Conversion reporting was the lowest-rated feature on the platform, and our largest client threatened to leave over it. Leadership’s first fix was to embed Google Looker. I pushed for a hybrid instead, and the dashboard it produced became the template for every dashboard at Madhive.

Role
Lead product designer, UX researcher
Scope
End-to-end design, strategy, usability testing, prototyping
Team
2 PMs, 10 engineers, 4 Looker BI engineers
Timeline
Feb 2024 to Jan 2025, released
1.2→4.5CSAT score
<15sper report, down from 3–20 min
100%client adoption

The deck

Walk through the project

The slides I present in interviews, embedded right here. The full version is available on request.

The problem

“We’ll leave if Madhive doesn’t fix latency and data issues of reporting.”

One of the largest clients on the platform

Conversion reporting is how station groups and agencies prove their campaigns work. It was the lowest-rated feature on the platform: a report took 3–20 minutes to build, then got about 10 seconds on screen.

Instead of using the dashboard, clients downloaded raw data and rebuilt it elsewhere. Leadership’s first fix was to embed Google Looker.

1.2 / 5CSAT before the redesign
3–20 minto build one report
Exportthe common workaround

The approach

Build vs. buy: I pushed back on “just embed Looker”

An iframe would have shipped fastest, but it would have rebuilt the same data dump inside our product. I made the trade-off visible before arguing for it.

  • Visualized the trade-off. Looker was fast, but it had no section-level filters.
  • Proposed a hybrid. Keep the Looker backend and put our own UI on top.
  • Tested before pitching. Four sessions with customer success, then leadership bought in.

Looker iframe

Option A: the conversion dashboard as an embedded Looker iframe, a long page of stock charts and tables

Fast to deploy, but no section-level filters and a stock look that didn’t match the platform.

What shipped

From data dump to analytical scaffold

Four principles turned a wall of numbers into a dashboard people read top to bottom, built in our design system on Looker data.

Select a card to see it larger.

A call I reversed

A pretty heatmap lost to a plain line chart

The heatmap looked great in mocks. With real client data, most cells landed in the same color band, so nothing stood out. The line chart showed the trend at a glance, so that is what shipped.

Explored: heatmapHeatmap of conversions by days since impression, most cells in similar shades
Shipped: line chartLine chart of conversions by days since impression, showing the drop-off clearly

Impact today

Became the dashboard template for the whole company

The suite is used by every client across 12,000+ campaigns. Our largest clients’ testimonials became part of the sales pitch, and every new dashboard at Madhive now starts from this one.

94%said it helped them find actionable stories (Sept 2026)
$6.4Mestimated operating cost saved by replacing manual reporting
50%of users moved over fully within two months

Want the full story? The complete deck covers research, explorations and rollout.
Email me for access: qinglu.design@gmail.com

Or discover inventive designs, creative solutions, and successful collaborations that demonstrate my knowledge and enthusiasm for AI, DataVis and more.