EMBROKER · FINTECH / INSURANCE · SELF-SERVICEReaching 93% self-service adoption for certificates
I redesigned an existing Certificate of Insurance request flow so customers could generate proof of coverage themselves instead of relying on Embroker’s service team.
The project in 20 seconds
Embroker is a digital insurance platform for businesses. A Certificate of Insurance (COI) is proof of active coverage, often requested by clients, partners, landlords, or vendors.
Certificate requests still depended heavily on Embroker’s service team, creating customer waiting and repetitive internal work.
Turn an assisted workflow into a clear, trustworthy self-service experience without removing access to human support.
Product design from problem definition and flow design through prototyping, testing, delivery, and post-launch iteration.
The result94% of eligible certificate requests became self-service in the first week.
Four weeks after launch, adoption was still at 93%, showing that self-service had become the default behavior, not just a launch spike.
Project details
Embroker
Business insurance platform
Certificate of Insurance self-service
Senior Product Designer
PM · Engineering · Customer Success
Shipped
Existing workflow redesign
A routine document with real business consequences.
A Certificate of Insurance sounds simple: proof that a business has active coverage. But generating one can involve recipient details, coverage requirements, policy rules, and cases that still need human review.
The challenge was to make the common case feel effortless without pretending every case could be automated.
Customers could request certificates from their account, but fulfillment still relied heavily on Embroker’s service team. Routine requests and genuinely complex cases entered the same operational flow, meaning customers often waited for a task that required little or no manual judgment.
The existing process worked, but it didn’t scale.
Customer costWaiting for a routine request
Customers depended on service availability even when their certificate required no special handling.
BUSINESS COSTRepetitive manual work
Service teams spent time processing standard requests that could be completed safely through the product.
PRODUCT OPPORTUNITYSeparate the common case from the exceptions
Automate what could be automated while keeping human support available where judgment was still required
My role
I owned the end-to-end UX for the self-service experience, from understanding the existing workflow through launch.
Product logic
Worked with product and insurance stakeholders to define what could safely become self-service.
Experience design
Mapped journeys and flows, designed the interaction model, and prototyped the end-to-end experience.
Validation and delivery
Tested key assumptions with users and supported implementation through design QA and launch.
Key product decisions
The hardest part wasn’t designing the form. It was deciding where users needed guidance, what could be automated, and where human support still belonged.
The simplest screen wasn’t necessarily the simplest experience
A certificate request can look like a short form from the outside. But users entered the process with very different levels of familiarity and confidence.
Mapping the journey showed that the biggest problem wasn’t simply the number of steps. It was uncertainty about what information was needed, what would happen next, and whether they were doing it correctly.
Journey mapping exposed where uncertainty peaked and where additional guidance would have the most value.Not every request should be automated
The goal wasn’t to remove the service team from every certificate request. It was to automate the common case while deliberately protecting the exceptions.
I mapped the branching logic with product and insurance stakeholders to define which requests could be completed independently and which still required human review.
The logic separated standard requests from cases that still needed service-team intervention.Less interface didn’t mean less uncertainty
I tested a lighter version of the experience against a more guided approach.
The lighter flow looked simpler, but users benefited from additional context at the moments where they were most likely to hesitate. The guided experience improved confidence from 8.7 to 9.3, so we kept the additional guidance instead of optimizing for fewer words and controls.
The flow could only reduce support dependency if customers found it. A standard banner risked being overlooked.
I designed a more noticeable feature announcement that stood apart from static page content. The goal was to make the new self-service option visible without making the experience feel disruptive.
Discovery was part of the product experience. A self-service feature has little value if users continue to rely on old support paths because they do not notice the new option.
Making the self-service path discoverable
Final solution
The shipped experience let customers start a certificate request from their account, understand what information was required, complete standard requests independently, and reach human support when their case fell outside the self-service path.
Clear entry point
Certificate requests became a visible part of the customer’s existing insurance experience.
Explicit review
Customers could confirm key information before submitting rather than wondering what would happen next.
Guided completion
Context and helper text appeared where users needed reassurance, without overloading straightforward steps.
Support for exception
Cases that couldn’t safely be automated still had a clear path to human help.
Design for desktop
1024
Design for smaller screens860776
Design for mobile
What happened after launch
Self-service adoption was immediate rather than gradual. 94% of eligible certificate requests were completed through self-service in the first week.
Four weeks after launch, adoption was still at 93%. The result wasn’t just a launch spike: customers continued choosing the self-service experience once it became part of their normal workflow.
Adoption held after launchImpact
94% self-served in week one
115 of 123 certificates were issued through the new flow.
8.7 → 9.3 confidence score
Testing showed higher confidence with the guided infobar version.
93% self-served across four weeks
Self-service adoption remained high after launch.
Reduced support dependency
Customers could complete a previously confusing certificate task independently.
Reflection
Self-service works best when it doesn’t pretend every case is simple. The common path should feel easy, while the product still provides clear boundaries for exceptions.
This project reinforced something I keep coming back to: making an experience feel simple for the customer often requires more structure behind the scenes.
Next project →From a freelancer pain point to a testable fintech concept in one day
I used AI to reduce the time between noticing a problem and creating something concrete enough to evaluate, then critically refined the output around financial logic, trust, explainability, and technical feasibility.