Designing a free, step-by-step guide that helps parents understand the IEP process and act on it, for Understood.org. From four concept prototypes to a shipped guide, then a reusable template the next guide was built on.
UnderstoodContext
Getting an IEP, an Individualized Education Program, is long, confusing, and full of barriers. Some parents give up on finding support. Others never start, because nobody told them they could. Watching a child fall behind brings up guilt, helplessness, and fear, and the guidance parents find is scattered across search, social media, professionals, and AI, with no trusted place to begin.
Understood had the credibility and the content. About one in five of its most visited pages were already about IEPs, 504 plans, or accommodations, and the two questions parents asked its AI assistant most were whether a child can get accommodations without an IEP, and what the difference between an IEP and a 504 is. What was missing was a path.
Discovery & Research
We focused on parents of kids aged 5 to 10 with learning and thinking differences, either curious about IEPs or already navigating one. They were also the largest group of parents on the site. Educators weren't the target, but we expected them to share it.
full list ✓✓✓✓
phased ✓
show progress ✓✓✓
one next step ✗
Concept testing: four prototypes, then two UI directions compared across four variables
Strategy & Framing
From the research, four jobs emerged. They became the structure of every section and step, and one rule sat above all of them: Understood doesn't decide whether an IEP is right for a child. No step is gated. Every step pairs something to learn with something to do.
A sequenced view of every phase and step, so a parent can see where they are, what's ahead, and that they can start the process themselves.
Each step carries the minimum a parent needs, in the format they prefer: read, watch, or listen.
Every step ends in something concrete: a template, a checklist, a question to ask at the meeting.
Lives on understood.org, reached from the articles parents already read, with the AI assistant one tap away. No account, no new app.
Design Process
The team moved from concept testing in spring to a launch timed for back-to-school in August 2026. I owned the design from the first prototype through launch and the iteration that followed.
The guide is a spine: sections for each phase of the process, numbered steps inside them, all expanded by default so the whole picture is visible. Parents can collapse sections, and the guide remembers their preference.
A Learn / Take action toggle lets a parent who only came for a form go straight to the templates, organized by the same sections. One structure, two ways in.
The step page went through the most rounds. Parents told us they feared missing information, and the accordion beat the carousel because the carousel hid more.
Each step became one to five drawers. Read is always there, with Watch or Listen as options. Then Take action, with templates and checklists. Then, when we had one, Hear from a parent.
Previous and next at the bottom, and a "Full guide" link that returns parents to where they were, so a step never feels like a dead end.
“I’m afraid I’ll miss something.”
Wanted the full list, not one next step.
Whole process, kept scannable.
The product brief's first requirement was that this would be the first of many guides, and it had to be reusable without engineering. So I designed the guide as a content model in the CMS: Hero, Sections, Steps, and a Step page made of Learn drawers, Take action, and Parent story.
Editorial guardrails kept it honest: required components, a required Read modality, at most two formats per drawer, character limits. An editor can build a whole guide without a designer or an engineer in the room.
The proof came when the 504 Guide for Parents was built on the same template with no new components. Next up are guides for diagnosis, MTSS, and finding a provider.
Key Design Decisions
We could have revealed the process one phase at a time. Testing said no: parents want to see the whole journey and decide where to go, and a guide that withholds steps reads as a guide that's steering. So every section is open by default, every step is reachable, and progress is a "Reviewed" check saved on the device with no login. After launch we added one guided way in, "Help me find where to start," with "Browse all steps" as the quieter option, because two equal buttons added a decision.
Content on the site was credible but static. Every step in the guide pairs the minimum knowledge with a concrete next action, so a parent never finishes reading without knowing what to do. The templates became the most-used part of the guide. Reference sheets led: Anatomy of an IEP, IEP vs. 504, Accommodations vs. modifications. When a step isn't quite your situation, a jump-ahead card at the bottom routes you to the one that is.
We gave parents Read, Watch, and Listen because testing said they valued the choice. Launch data added a twist: audio performed like video when it sat in the drawer that opens by default, and media placed lower on the page barely got used. Where a format sits matters more than which format it is. That reshaped how we lay out step pages.
Outcomes
The IEP Guide launched on understood.org on August 10, 2026, in time for back-to-school: its own landing page, the full guide view, step pages, downloadable templates, on-device progress, and entry points from the articles parents already read. By October it was on track for its 2026 behavior-change goal, and the 504 Guide had shipped on the same template.
Reflection
Launch taught us more than testing had. Reach fell off steeply after the first section, and half of the test parents never left it. One thought he'd finished. Parents further along landed on early steps and didn't find what they needed. More than half of visitors were on desktop, and we had tested only on mobile. Five of twelve parents asked, unprompted, for a printable version of the whole guide to bring to a meeting. Each of those became the next round of work.
That next round is where my process changed. Research needed realistic, hosted variants of the live guide for unmoderated testing, and the framings we wanted to test didn't fit the CMS model. So I built the prototypes in code with Claude, on the real design tokens, fonts, and brand art, each variant a data file rather than a new set of components, on neutral URLs so participants couldn't tell which one they had. I rebuilt the control as a prototype too, so the test compared framing rather than polish. Changes went out the same day after each round.
The wayfinding study came out of that. Parents wanted one guided start, with browsing as a quieter option. "Find my step" got nearly everyone to a relevant step but overpromised an exact match. Jump-ahead beat the sidebar. Better navigation also made the gaps in content depth easier to see. That became v2.0: one primary call to action, jump-ahead at the end of every step, the sidebar deprioritized. The same shift ran through the brand refresh and the design system: a token pipeline from Figma to code with an accessibility gate, a hero system with every color combination checked to AAA, and dark mode across the site.
The pivot moment came after months of slow, manual Figma work, when I asked: what if I didn't have to do this by hand? My judgment drives each iteration. The hands-on work is where the tooling changed. I gave a company-wide talk on that shift.