Shaping Attentive's segment builder to match how marketers think
Attentive is a conversational-commerce platform that helps brands reach customers through personalized text messages. Its segment builder lets marketers create subscriber segments from behavioral and demographic data to target high-intent subscribers with personalized campaigns.
The segmentation criteria were hard to find and harder to understand. As a designer on Attentive's Targeting team, I led the redesign of the segment builder's information architecture, reorganizing top-level conditions to match how marketers actually think about their customers and exposing the data sources behind each one.
Segmentation criteria were nearly impossible to find
Prior research showed that Attentive users had trouble locating and understanding the segmentation criteria in the segment builder. When tasked with creating a segment from a specific type of data, users struggled to find the right condition in the dropdown menu, and often asked Client Strategists for clarification on existing conditions. The menu's lack of organization kept users from knowing which segmentation features were available, and created extra work for Client Strategy and support teams.
A chance to build for how marketers actually think
I hypothesized that a clear information architecture framework would build familiarity and predictability for users, increasing discoverability and efficiency during segment creation. This project was also a chance to build a framework for how data is surfaced in the builder, which would save time when adding new options later.
I planned to measure success by comparing the number of segments created by users versus Attentive support, and by reducing the volume of questions coming into support.
The original dropdown menu had 26 options, with more being added every month.
Grounding the redesign in research and collaboration
The new categorization had to reflect how marketers actually think, not how the product was built internally. Here's how I got there:
- Audited the segment builder's existing terms and conditions
- Ran card sorting exercises with users
- Partnered with the Design Systems team to build a new design system and components
- Met regularly with Content Strategy to review copy
- Validated concepts via unmoderated user testing and stress-tested segment conditions with cross-functional partners
Early design explorations
A segment builder built around real mental models
I used insights from discovery research and card sorting to inform the segment builder's new information architecture, then updated its visual hierarchy to match.
Three top-level categories
During card sorting, participants grouped segment builder options based on their own workflows and how they viewed their customers' journeys. I structured the new information architecture around those mental models, grouping every condition into three top-level categories: "A subscriber's activity," "A subscriber's characteristics," and "The segment a subscriber is in."
Visible data sources and definitions
Clients regularly asked Client Strategists to clarify existing data conditions, and user research surfaced the same confusion. Several participants asked how options were calculated, and understanding that changed how they approached building a segment. I designed a new menu component that exposes each condition's data source and definition directly in the builder, so users could build segments with confidence.
A new visual hierarchy
A brand refresh landed mid-project, giving me the chance to redesign the segment builder's visual treatment alongside the IA work. I used input size, proximity, alignment, and repetition to build a consistent, predictable layout with distinct groupings for each condition. Now, top-level condition inputs sit on their own line, with subsequent selections grouped below. I also moved the filter button to match our users' reading path.
Increasing user-created segments
The redesigned segment builder shipped in June 2021, giving users a more seamless and intuitive flow. It was received well — several clients emailed in, excited to discover previously-shipped features for the first time simply because of the filter button's new placement.
- Average segments created per customer increased by 15% post-launch, meaning fewer segments had to be built by support teams. This directly contributed to the company goal of creating a more self-serve product.
- The new information architecture framework reduced the time our team spent adding new conditions to the segment builder in later projects.
What our team would do differently
After completing this project, our team held a retro to reflect on what went well, what slowed us down, and where we could realign on our long-term vision for the segment builder.
- This was a large project that could've been broken into smaller ones. Our team agreed to start using project sizing and to align on clear deadlines before kickoffs going forward.
- With thin Content Strategy resourcing, terms and definitions across the product could easily drift. We agreed to work with other teams on a process for developing and maintaining consistent content.
- We wanted to share in-progress work earlier and more often so stakeholders had visibility sooner. Ideas included Gantt charts, Slack updates, and team demo days.
What I learned about advocating for design work
This was one of the first large projects I led and advocated for from start to finish. My manager, product partner, and I spent time making the case for why it mattered before it was prioritized. Once it was, I kept stakeholders aligned with regular research reviews — walking through findings, design direction, and next steps, and taking their questions directly. Advocating for this project taught me that a good pitch isn't enough. You have to keep making the case for why the work matters and keep stakeholders bought in all the way through.
Our team loved the sailboat retro format.