SEGMENT ANALYSIS SERVICE

Enterprise Data Visualization and Decision Support - Turning Complex Enrollment Data Into Actionable Decisions


Role: Senior Product Designer; end-to-end product ownership

Focus: Data Visualization, Information Design, Enterprise UX, Accessibility

Product: B2B SaaS / Enterprise Analytics

Users: College and university enrollment teams

Team: Product, Engineering, QA, UX Research, Content Strategy, Accessibility and stakeholders

Tools: Figma, FigJam, enterprise design system, user interviews

Timeline: 5 weeks

Redesigning College Board’s Segment Analysis Service to help enrollment professionals upload institutional data, interpret student segments, compare performance, and act on insights with greater confidence.


The product

Segment Analysis Service helps colleges analyze their enrollment data through College Board’s student-segmentation framework. Institutions upload student records, receive enriched cluster data, and use reports to understand which student groups are progressing from prospect to enrollment.

My role

As Senior Product Designer, I led the experience and visual design across the upload, processing, reporting, and accessibility workflows. I translated complex analytical requirements into an understandable interface while defining reusable patterns for filters, chart interactions, tables, system feedback, and accessible color presentation.

  • End-to-end workflow and information architecture

  • Dashboard and report hierarchy

  • Data-visualization design

  • Filter and comparison patterns

  • Accessibility modes and chart-color testing

  • Upload, validation, confirmation, and processing states

  • Collaboration with product, engineering, analytics, and accessibility partners

  • Reusable interaction patterns for future analytics products

Users and decisions

I designed around the decisions users needed to make—not around the structure of the underlying dataset.

  • Enrollment leaders evaluating recruitment strategy

  • Admissions analysts investigating student populations

  • Institutional users preparing data for analysis

  • Staff translating report findings into student-search criteria

Questions the experience needed to answer

  • Which student segments are strongest or weakest?

  • Where do prospects drop off between applicant, admit, and enroll?

  • How do results differ across cohorts or uploaded files?

  • Which clusters should be targeted in future recruitment?

  • Can users trust that the uploaded information was processed correctly?

This shifted the product from “generate a report” toward a decision-support experience: select the correct evidence, understand what matters, compare performance, and move the insight into an actionable recruitment workflow.

Experience strategy

I organized the experience to match the user’s mental model rather than exposing every analytical capability at once. New users receive orientation and guidance, while returning users can move directly into uploads, cluster definitions, or historical reports.




  1. GET STARTED
    Understanding the service
    and available actions.

















  2. UPLOAD FILE
    User prepare, upload, validate, and
    submit institutional data.

    Users needed to prepare and submit institutional data before they
    could generate a report. I designed a guided upload process that
    explains the available options, provides resources when needed,
    confirms the file was uploaded successfully, and lets users
    review everything before submission.

    • Clear explanations for historic and periodic tagging

    • File templates and preparation instructions

    • Required and optional fields

    • Upload success and processing messages

    • A review screen before final submission

    • Clear next steps after submission

    • The new workflow helps prevent upload mistakes and keeps
      users informed throughout the process.

































  3. CLUSTER GUIDE
    Understand the meaning behind each
    segment.














  4. REPORTS
    Filter, compare, interpret, download
    and apply results.



Making the Data Accessible and Trustworthy

Accessibility was part of the information design—not an afterthought
With multiple student clusters displayed at once, relying on color alone could make the report difficult to interpret and increase the risk of users reaching the wrong conclusion.

I designed an accessibility option that adjusted the color palette while preserving the relationship between the chart, legend, and table. Cluster names, exact values, headings, and positional patterns remained visible so users did not have to depend entirely on color recognition.

This approach improved the experience for users with visual disabilities while also making the report clearer for anyone reviewing dense or unfamiliar data.

Accessibility principles applied

  • Color was supported by labels and numerical values

  • Categories remained consistent across charts, legends, and tables

  • Selected, disabled, and focus states remained visible

  • Supporting tables provided an exact-value alternative

  • Keyboard behavior was considered across filters and actions

  • Information remained understandable without color alone

Accessibility statement

Designed with accessibility guidance and evaluated against applicable accessibility standards. Formal WCAG compliance would require a complete implementation audit.


The Challenge


Accurate data was available, but understanding and acting on it required too much interpretation.

Enrollment professionals needed to move between data preparation, uploaded files, static cluster documentation, historical reports, and student-search tools. The experience placed a high cognitive burden on users and did not clearly communicate what was required, what the data represented, or what action should follow.

Core problems

  • Users needed specialized knowledge to interpret cluster codes.

  • Required report criteria were not clearly separated from optional filters.

  • Visualizations contained many categories competing for attention.

  • Users needed to compare multiple stages of the enrollment funnel.

  • Report findings had to connect back to student recruitment workflows.

  • Uploading and processing data lacked sufficient visibility and reassurance.

  • Color-dependent reporting created accessibility and comprehension risks.


Applying Human-Centered Visualization Principles

The system was designed around how people scan, compare, and understand information

I used visual hierarchy to direct attention toward the most important information first. Position and bar length supported quick comparison, while color and grouping helped users recognize related categories.

Filters were grouped by purpose, report controls remained visually separate from results, and repeated patterns behaved consistently throughout the workflow. Optional complexity was introduced only after users completed the required report setup.

Exact labels and values remained available alongside the visualization so users could confirm what they were seeing. This helped balance quick pattern recognition with the accuracy and credibility expected from an enterprise analytics product.

The design principles

  • Hierarchy: Lead with the primary question and most important result

  • Comparison: Use stable positioning and scale across enrollment stages

  • Grouping: Keep related filters, controls, and report actions together

  • Consistency: Preserve the same category mapping throughout the experience

  • Progressive disclosure: Introduce advanced options only when relevant

  • Trust: Let users verify every visual pattern through supporting values


Creating Reusable Visualization Standards

The project established a foundation that could extend beyond one report
As lead visual designer, I used this work to advocate for stronger data-visualization capabilities across College Board products.

The existing design system supported general interface patterns but offered limited guidance for analytical experiences. This project identified the need for expanded color options, reusable graph components, accessible palettes, standardized legends, report states, and consistent filter behavior.

I worked to make these patterns reusable so future product and analytics teams would not have to solve the same visualization problems independently.

PatternStandard establishedReport setupRequired inputs appear before optional controlsFiltersConsistent counts, update, reset, and clear behaviorEmpty statesExplain what is missing and what users should do nextChart colorsStable category mapping with accessible alternativesLegendsConsistent placement, labels, and chart relationshipsData tablesExact counts and percentages support visual interpretationReport actionsConsistent placement for generation, download, and next stepsSystem feedbackShared loading, success, processing, and error statesGraph componentsReusable chart structures designed for future products

This expanded my role beyond designing a single report. I became a voice for visual design across the organization—helping teams understand when visualization was useful, how color could communicate meaning, and how reusable graph patterns could broaden what the design system offered.

Leading Across Product, Analytics, and Engineering

Translating analytical requirements into patterns teams could implement consistently

I partnered with product, analytics, engineering, accessibility, and design-system teams to balance data accuracy with user comprehension.

My role was not to determine the underlying analysis. My responsibility was to understand what the data represented, clarify which comparisons were meaningful, and design the clearest way for users to interpret and act on those findings.

I led decisions around:

  • Report hierarchy and visual storytelling

  • Chart structure and category encoding

  • Valid comparison and aggregation states

  • Loading, empty, error, and processing experiences

  • Accessible color and interaction behavior

  • Reusable graph and filter components

  • Expansion of the design system’s visualization options

  • Handoff guidance for future analytical products

This work helped bridge the gap between technically accurate data and an experience users could understand, trust, and use.

Measuring Success

Success meant users could configure, interpret, and act on the report

The experience was evaluated around three questions:

  1. Could users build a valid report without unnecessary assistance?

  2. Could they correctly interpret the most important patterns?

  3. Could they turn those findings into a recruitment action?

  • 10% increase in successful report configuration

  • 9% decrease in time required to generate a valid report

  • 12% decrease in file-upload or configuration errors

  • 29% improvement in identifying top- and bottom-performing clusters

  • 23% increase in reports leading to saved recruitment criteria

  • 33% reduction in report-related support requests

  • 30+ reusable visualization patterns introduced or identified

  • 15 College Board experiences positioned to adopt the new patterns

Impact

From a static report to a connected decision-support experience

The redesign connected data preparation, report configuration, visualization, verification, and recruitment action within one coherent experience.

Clearer decisions

Users could recognize important patterns without manually interpreting a dense report.

More trustworthy information

Charts, direct labels, legends, and tables allowed users to scan the data and verify exact values.

More inclusive reporting

Accessible palettes and redundant visual cues reduced dependence on color alone.

More actionable insights

Users could move directly from identifying a student cluster to adding it to recruitment criteria.

A scalable visualization foundation

The work established patterns for charts, filters, legends, tables, accessibility, and report states that could expand into other College Board products.

The broader impact was not simply introducing a new graph. It demonstrated how human-centered visualization could make complex enterprise information easier to understand and helped move College Board toward a more capable, colorful, and reusable analytics design system.

Reflection


Good visualization does not simply display information—it helps someone make the right decision

This project reinforced that data-visualization design begins before the chart and continues after it.

Users need confidence in the source of the data, clarity about the filters being applied, an accessible way to interpret the result, and an obvious path toward action. If any of those pieces are missing, an accurate visualization can still fail its users.

As lead designer, I also learned that one successful visualization can create momentum beyond a single product. By advocating for expanded color guidance, graph components, accessible patterns, and stronger visualization standards, I helped create a foundation that other College Board teams could build upon.

In the next iteration, I would test clearer funnel-conversion indicators, reduce simultaneous color complexity, evaluate additional comparison formats, and formalize the strongest patterns into a complete enterprise visualization system.

This work transformed complex enrollment data into an accessible and actionable decision-making tool—and established a stronger foundation for how visual data could be communicated across College Board.







Designing Data for Faster Decisions


I designed the report to help enrollment teams quickly understand which student segments were performing well, where students were dropping out of the enrollment funnel, and where recruitment strategies could be adjusted.

Instead of presenting users with a dense report, I created three levels of information:

  1. See the pattern through the visualization

  2. Verify the numbers in the supporting table

  3. Act on the insight by adding selected clusters to recruitment criteria

The stages—Suspect, Prospect, Applicant, Admit, and Enroll—remain in a consistent order so users can follow movement through the funnel. Direct labels, stable positioning, and consistent cluster colors reduce the effort required to interpret the chart.

The result was more than a visualization. It connected data interpretation directly to the next business decision.

Previous
Previous

SAT Suite Score Reporting

Next
Next

Operational Management Platform