Cordial · Audience segmentation case study

Audience Builder

Visual audience segmentation for Cordial's marketers

Cordial's segments lived in a JSON editor that only advanced users could work in. I led design on a drag-and-drop builder any marketer could use. After launch, per Cordial data, segment creation got roughly 40% faster.

Role
Design lead, Audience BuilderOfficial title: Senior User Experience Designer
Timeline
10 to 12 monthsAt Cordial, 2017 to 2020
Research method
Interviews and testing4 to 5 participants, internal teams and clients
Segment creation time
Roughly 40% fasterThan with the old JSON editor

The problem

All audiences meant JSON. Marketers waited for help.

Marketers wanted more control over who got each email, SMS and in-app message. Rules and rule sets lived in an old code editor built for people fluent in JSON. Here is the same audience both ways.

BeforeWritten by hand in JSON

Only people fluent in JSON could change it.

AfterBuilt on the canvas

Any marketer can drag it together.

What research found

Three problems kept coming up.

JSON only

Only JSON experts could build rules.

Custom rules needed JSON knowledge, so marketers leaned on customer support and waited for the turnaround.

New rules waited on support

No nesting

Saved audiences couldn't be reused.

There was no way to nest a saved audience inside a new one, which made advanced campaign logic harder to build.

No way to build on a saved audience

Scale

Big lists were slow to segment.

The old setup couldn't handle large queries, so segmenting big contact lists meant delays for campaigns.

Querying

Big lists meant delays

What we set out to do

We set three goals for Audience Builder.

Marketers of every skill level should build and reuse audiences on their own, without retraining.

  1. Goal 1

    No JSON required

    A visual canvas where you drag or click rules into place, so anyone can build a segment.

  2. Goal 2

    Reuse saved audiences safely

    Nest a saved audience inside a new one without changing the original or relearning anything.

  3. Goal 3

    Stay fast on big lists

    Large data sets shouldn't slow targeting down on email, SMS or in-app.

My role: Design lead on Audience Builder at Cordial, where my official title was Senior User Experience Designer. Over 10 to 12 months I took it from research to UI designs and prototypes, working closely with the PM, subject matter experts (SMEs) and backend developers.

Team
1 PM, 3 SMEs and 5 backend developers (2 onshore, 3 offshore)
What I owned
Competitor analysis, user studies, personas, wireframes, UI designs and prototypes
Mentoring
I helped 2 designers grow through regular feedback and a weekly critique I started.
  • Competitor analysis
  • User studies
  • Personas
  • Empathy mapping
  • Wireframes
  • Prototyping
  • Drag and drop
  • Marketing automation

The 30-second read

Segments moved off code onto a canvas.

Segments took roughly 40% less time to build than with JSON.

01 · What I designed

Drag a rule in,
or click to add it.

Include and exclude groups, rule sets and saved audiences all live on one canvas. The code editor is still there for experts.

02 · What changed

Marketers built their
own segments.

After launch, per Cordial data, support tickets fell by roughly 55%.

Walkthrough · 45 seconds

You drag a rule in and drop it where it goes.

The pink marker shows where each rule will land as I drag.

Research

I designed for three kinds of marketers, new to expert.

User interviews and the time customer support spent on manual work shaped the direction.

Beginner user

New Marketer

Design influence: Pointed to friendly visual aids and simple, short workflows.

Way inDrag a rule

Intermediate user

Mid-Level Marketer

Design influence: Shaped a mix of visual tools and code, with fixes for common errors.

Way in</> Visual and code

Experienced user

Senior Marketer

Design influence: Pushed for quick linking and real-time hints, without risking the platform's stability.

Way inSaved audiences

Full persona details

New Marketer (beginner user)

  • Demographics. Entry-level marketer, 22 to 30, a recent grad or one to three years in, new to Cordial, runs basic campaigns for small audiences.
  • Goals and needs. Quickly create simple segments for promos or cart abandonment with visual tools and as little scripting as possible.
  • Pain points. Overwhelmed by complex queries, makes frequent grouping errors and relies on help for changes.
  • Behaviors. Enjoys drag and drop and prefers a more automated approach.

Mid-Level Marketer (intermediate user)

  • Demographics. Mid-career specialist, 30 to 40, four to seven years in digital marketing, an average Cordial user handling mid-scale targeting.
  • Goals and needs. Builds moderately layered segments with instant edits and fast previews, across channels.
  • Pain points. Slowdowns from outdated systems, and trouble linking segments without changing campaigns.
  • Behaviors. Mixes visual tweaks with basic coding and values speed in team edits.

Senior Marketer (experienced user)

  • Demographics. Seasoned marketer, 40 to 55, eight or more years in marketing ops, a heavy Cordial user who runs big multi-group campaigns.
  • Goals and needs. Build and save complex queries with flexible links, with safe edits and smooth channel integration.
  • Pain points. Few ways to connect segments, campaigns split across platforms, and edits that risk existing setups.
  • Behaviors. Relies on keyboard shortcuts, cares about precision and scale, and often links segments for promotions.
Competitor analysis

I reviewed the tools our marketers compared us with.

In research and exploration I looked at Mailchimp, Marketo, Braze and Iterable. They left gaps in scale and in nesting saved audiences, which Audience Builder set out to fill.

  • People wanted it visual.Non-technical users liked Mailchimp for its simplicity and wanted a hands‑on builder with drag and drop.
  • They wanted to reorder and nest.Users asked for a quick way to reorder segments and for better nesting for dynamic campaigns.
What I found in each tool
Small business email

Mailchimp

It led small-business email automation with an accessible UI, but advanced segmentation struggled with larger audiences.

Mid and large B2B

Marketo

A preferred choice for mid and large B2B companies with strong capabilities, but a clunky UI for non-technical users and limited real‑time nesting.

Cross-channel

Braze

Known as a cross-channel leader, with higher costs and a steeper learning curve, and limits on nested groups and performance on big data sets.

Multi-channel

Iterable

Timely multi-channel targeting, but restrictions on nested audience groups and backend performance for larger data sets.

Empathy map for the three personas

We ran an empathy map exercise to explore what each persona needed, what frustrated them and what drove them, so the design matched their real work.

Empathy map for the three Audience Builder personas, with what each one says, thinks, does and feels.

Design

Four decisions, from the canvas to the code editor.

The canvas sits on the left and the rule picker on the right.

Early wireframes (6)

In early exploration we showed low-fidelity wireframes to stakeholders and clients. Their feedback on the drag-and-drop ideas shaped the high-fidelity designs.

01 · Drag or click

Drag a rule onto the canvas, or click it.

Every rule in the picker can be dragged into Include or Exclude, or clicked to add it where the blue highlight is. While you drag, a pink marker shows exactly where it will land, and the other group shows a gray one.

AnswersOnly JSON experts could build rules

02 · Rule sets

Make a rule set two ways.

Drop a rule onto another rule, or drag a Rule set in from the picker and fill it. In testing I asked people whether both ways were easy to find.

Tested inThe prototype sessions

03 · Saved audiences

Nest a saved audience inside a new one.

Saved rules holds the audiences your team has already built. Drag one in to build on it. If someone edits the original, a pop-up highlights what changed and lets you use the update or keep your version.

AnswersSaved audiences couldn't be reused

04 · Code editor

Keep the code editor for experts.

Early studies showed some technical users wanted to keep writing JSON. Some wrote it in an outside code editor and pasted it in. We upgraded the JSON editor with real code editor features, and dropping that outside step was a big win for them. Non-technical users preferred the visual builder.

Shaped byThe Mid-Level Marketer

  • Syntax highlighting
  • Line numbers
  • Autocomplete
  • Error highlighting
  • Bracket matching
  • Code folding
  • Keyboard shortcuts

Under the hood

Big lists and old segments were engineering work too.

I worked on both with the five backend developers on the team.

Large audiences

Big contact lists had to stay fast.

Scaling large queries was backend work. On the design side, rules are edited right on the canvas, which cut down the modals people had to open and close.

Big lists, fewer modals

Backward compatibility

Existing setups had to keep working.

Every existing JSON audience had to open in the new builder. That was a project requirement, so people wouldn't need retraining. The code editor stayed one click away on the toolbar for anyone who preferred JSON.

Both ways stay open

Testing

I tested with internal teams and client groups.

I wanted to see if people could build a segment and reorder its rules on their own, and whether anything felt confusing.

Who took part
Internal teams and client groups, 4 to 5 participants, in rounds of testing plus live reviews for UI tweaks. I recorded sessions with a drag-and-drop prototype to see how people used it.
What they said
Strong approval. Testers called the setup "smooth yet powerful" and built simple segments quickly. Self-service meant fewer support tickets, and people got more control over their own segments.
Questions I asked

I tested the decisions above, one by one.

  1. Did the drag-and-drop feel clear and responsive? If not, what felt confusing or difficult?
  2. Was the placement indicator clear when you reordered a rule?
  3. Were the labels "Include Contacts" and "Exclude Contacts" clear and descriptive?
  4. Was it clear you could create a rule set in two ways? Now that you know, are both easy to find?
All ten questions from the sessions
  1. What is your first impression of the layout, and how easy does it feel to get around?
  2. Can you walk me through how you would create a new audience segment with this prototype?
  3. Did the drag-and-drop feel clear and responsive? If not, what felt confusing or difficult?
  4. If you add several rules and want to reorder them, how would you do it?
  5. Was the placement indicator clear when you reordered a rule?
  6. How easy was it to find and pick specific rules or saved audiences in the right panel?
  7. Were the labels "Include Contacts" and "Exclude Contacts" clear and descriptive?
  8. How would you create a rule set? Was it clear you could create one in two ways? Now that you know, are both easy to find?
  9. How would you save a segment? Did anything feel missing or unclear?
  10. If you could change one thing about this prototype, what would it be?

Results

Segment builds got faster, and support tickets fell.

Opening segment building to every skill level led to a rise in audiences created and wider use across Cordial's platform. All numbers are after launch, per Cordial data from product analytics and support ticket counts.

Segment creation time
Roughly 40% less time to create a segment
Support tickets

Roughly55%fewer support tickets

Self-service cut the wait for custom segments.

Targeting

Per Cordial dataSharpertargeting

One segment worked across every channel.

Adoption

After launchGrewmore segments built

People praised nesting and the interactive UI.

Looking back

What held up, and what I'd A/B test.

Sessions had 4 to 5 people, so the first numbers came after launch.

What held up

Working with backend developers changed the design.

Tackling outdated components, compatibility issues and nesting together is what moved rule editing onto the canvas, with fewer modals in the way.

What I would change

I would A/B test the UI earlier, and more broadly.

In retrospect, early A/B tests would have measured how the UI affected campaign speed, on top of the testing rounds and live reviews we ran.

Audience Builder · Cordial

Thanks for reading.

I moved segment building out of a JSON editor and onto a canvas, so any marketer at Cordial could build and reuse an audience.