Herald API

UX Research & Product Design

Herald's API had a conversion problem. Brokers were interested, but their developers couldn't integrate quickly enough to close deals. The documentation was text-heavy, the site had dead ends, and there was no visible proof that any of it worked. I spent six months working with their team fixing all three.

Role

Tools

Type

Timeline

Product Design Intern

Figma, Claude

Client Work

6 months

The Challenge

The documentation was just one part of the problem. The site itself had navigation issues that left users stuck with no way back, and there was no social proof anywhere to convince potential partners that Herald actually worked. I tackled all three: the docs, the site UX, and the missing case study and partner showcase pages.

Documentation Design

Herald's documentation was all text. If a developer was trying to figure out how Webhooks worked or when to use polling, there was nothing to look at, just walls of copy to read through.

I worked with my mentor to actually learn how the API worked first, starting from scratch with no technical background. His explanation of Webhooks to someone who had never heard of them became my design brief. If I could follow a diagram well enough to understand it, a developer with real context would find it even clearer. I used that as my bar for every visual I made: Webhook overviews, polling comparisons, usage flows, and integration diagrams.

I built a rapid iteration workflow using Google Stitch and Claude to generate multiple visual directions from prompts, then evaluated and refined the strongest options in Figma. What would have taken 3 to 4 days of sit-down work took about 4 to 6 hours total. That freed me up to focus on the decisions that actually mattered: which diagram structure would work for a developer trying to debug an integration.

Herald API Website Audit

The site had more problems than just missing social proof. I ran a full audit and found issues across navigation, accessibility, and content that were quietly frustrating users and killing conversions. The biggest ones:

The Insurance API Index page stripped out the main nav entirely, leaving users with no way to get back
Button styles and capitalization were inconsistent across pages, making the site feel unfinished
Low-contrast CTAs failed WCAG standards, meaning a chunk of users couldn't even read them clearly
The Partnerships page had no CTA at the bottom, so anyone who made it to the end had to scroll all the way back up to take action

Competitor Analysis

I looked at how three comparable companies, Bundl, Limit, and Sayata, handled social proof. All three led with evidence: real client quotes, named companies, specific reasons why they were the better choice. Herald's site was doing the opposite, lots of "we can do this" with nothing to show for it. The gap was obvious.

Designing the case study and partner showcase pages was straightforward because Herald already had a solid design system in place. I worked within their existing components, matched their tone and language, and kept everything consistent with what was already there. The goal was for the pages to feel like they had always existed, not like something an intern threw in.

Outcome

4-6 hours

75→90%

4

What used to take
3 to 4 days of iteration

WCAG compliance after

accessibility fixes

New page designs

delivered in Figma

Case study page framework and partner showcase.

Designed in Figma

Reflection

Working inside a startup taught me how much design depends on the people around you. Having a design system already in place made the work feel like plug and play, and that was a real surprise. I came in expecting friction and found flow instead. If I could go back, I'd ask a lot more questions, about the developers, the users, the roadmap. I was juggling a full course load and a job at the same time, and I know I left some things on the table because of it.

What I Want Next

Herald reminded me why I got into design. I want to be in a small, scrappy team where I can do work that actually helps people, whether that's making a developer's life easier, cleaning up something that's been broken for years, or just making something that didn't look right finally look right.