Gregory was the first and sole designer at Chattermill for two years. He redesigned every screen: onboarding, feedback ingestion, analytics, dashboards, workflow automation, and settings. He built the design system from scratch and mapped every token to frontend code.

Research drove the work. He ran customer interviews, UEQ surveys, and session recordings. He instrumented features with Segment + Fullstory so new releases had analytics from day one. The research practices he set up stayed after he left.

NPS went from 53 to 75. Enterprise prospects started citing the interface quality in demos. Support complaints dropped. Every UEQ dimension improved. The redesign helped close the Series B.

2019 – 2021

Chattermill turns customer feedback into structured analytics. Text, audio, video reviews go in. NLP categorizes them. Charts and dashboards come out. I joined as the first designer to redesign the whole platform.

Full platform redesign Design system User research Enterprise UX NPS 53 → 75
Chattermill platform overview

Focus

The existing app was built fast. It worked, but usability, visuals, and core flows all needed rethinking.

I started by collecting pain points and feedback, then designed an approach where old features could coexist with new ones during the rollout. The visual quality mattered too. In enterprise analytics, most competitors look the same. A polished interface was a real differentiator in demos.

Interface scalability

Reusable components, design system, guidelines.

Aesthetics & accessibility

Modern UI with accessible color palettes for data visualization.

Self-serve onboarding

Step-by-step integration setup and data source connection.

Flexible reporting

Custom dashboards, report builder, and workflow automation.

Scope

"Redesign" is vague. I broke it into stages, each with specific steps, so the team could see what was happening and when.

Observation

  • Watching user sessions
  • Working in the support
  • Structuring complaints and requests
  • Analyzing competitors

Testing & improvement

  • Watching user sessions
  • Collecting feedback
  • Mechanics improvements and visual restyling

Implementation

  • Designing high-level UI concepts
  • Layout and basic elements design
  • Functional components and interactions
  • Corner cases
  • Assembling design system
  • Writing specifications

Hypothesis validation

  • Quick surveys
  • Event tracking
  • Customer development sessions
  • Determining missing backend
  • Prioritization and estimation
  • Defining the scope of the first versions

User research

Qualitative

Customer development interviews, UI feedback calls, session recordings, and every support request that came in. I read all of them.

Quantitative

UEQ surveys before and after the redesign, event tracking, and quick unmoderated surveys. I used Segment.io's visual tagger to instrument new features and piped events to Fullstory. New features had analytics from day one.

Value proposition canvas workshop

Onboarding

Challenge

Every new customer needed the Chattermill team to configure the platform manually. Nobody could get started on their own.

Solution

A self-serve onboarding flow: signup, pick your integrations, connect data sources. Customers saw their first results without waiting for anyone on our side.

Aggregating feedback

Feedback comes in from dozens of channels: surveys, app reviews, support tickets, social media. I designed the interface for pulling it all into one place, categorizing it with NLP, and showing what actually matters.

Building reports

The Analytics page is a report builder. Group metrics by attributes, apply filters, pick a chart type. The data updates live.

Impact of themes

The Analytics page also has predefined reports. Impact of themes shows which topics in customer reviews drive sentiment up or down, so teams can see what's helping and what's hurting.

Assembling dashboards

Customers can group any reports into shareable dashboards. Drag widgets in, rearrange, share with the team. One screen shows feedback trends and individual reviews side by side.

The hard part was keeping dashboards readable. Too many charts become noise. I designed the layout so each report reads clearly on its own but the full dashboard tells a coherent story.

Dashboard assembly overview

Adaptive widgets

The old app had two widget sizes and just stretched chart content proportionally. I redesigned it so charts adapt their content to the widget size: a small widget shows a summary, a large one shows the full breakdown.

Workflow automation

Seeing a problem in the data is one thing. Doing something about it is another. I designed workflows that trigger actions automatically: send a Slack alert when NPS drops, create a Jira ticket when a theme spikes, push data to a CRM.

Workflow builder

Each workflow has a trigger, conditions, and actions. Users pick a feedback pattern, set thresholds, and choose what happens next. The builder shows the full chain so nothing fires by surprise.

Settings & administration

I restructured settings around how teams actually manage their workspace: integrations, user roles, data sources, billing. Fewer clicks to find what you need.

Outcome

The redesign shipped and the numbers moved. UEQ scores improved across every dimension (scale: −3 to +3). Business metrics followed.

Analysts got saved filters and report templates one click away. Managers could pin dashboards, share them externally, or embed them in Notion.

Review tagging went from manual to mostly automated, and self-serve onboarding replaced live training sessions.

1 2 3 Attractiveness 1.08 → 2.41 Perspicuity 1.15 → 2.33 Efficiency 1.28 → 2.19 Dependability 1.19 → 1.73 Stimulation 0.99 → 1.80 Novelty 0.68 → 2.11
53→75 NPS jumped

Net Promoter Score went from 53 to 75 after the redesign shipped.

Better demos

Enterprise prospects started commenting on the interface quality during demos. Conversion went up.

Fewer support tickets

Complaints and bug reports to support dropped sharply after launch.

𝙱 Series B closed

The redesigned product helped close the Series B round.