For one client platform, accessibility had been considered across the core application, but the account setup experience had grown over time. Additional questions had been added, instructions had become longer, validation rules had become more complex, and guided overlays had been layered on top of the original flow.
What looked internally like a sequence of onboarding screens was experienced by users as one long task requiring them to read, remember, enter, correct, navigate, and repeatedly reorient themselves. The number of new users who would abandon the platform midway increased with every new release - on its worst release, 55% of users who started onboarding did not complete the setup.
The Challenge
The onboarding journey had become more demanding than the product experience it was supposed to introduce.
Users had to complete eight separate onboarding steps before gaining full access to the application. Some information was essential for account creation, while others were being collected because they might be useful later. To the user, however, they all appeared equally mandatory.
That additional friction was showing up in the data:
- 26% of users abandoned onboarding before activation
- 43% of onboarding sessions involved at least one validation retry
- 35% of keyboard-only test participants required assistance to complete the journey
- Median onboarding time had grown to approximately 11 minutes
- Users had to complete eight steps before reaching the core product
The numbers pointed to a broader problem. No single screen accounted for the drop-off. Friction was accumulating across the entire journey.
There was a cognitive burden. Users had to process long instructions, remember requirements from previous screens, make decisions without always understanding why information was being requested, and maintain their place across a process whose length was unclear.
There was also a form interaction problem. Password rules, accepted formats, and required information were not always communicated before users interacted with a field. Users often discovered requirements only after validation failed.
For screen-reader users, dynamic validation created another layer of difficulty. An error could be announced without making it immediately clear which field caused it or what action was required.
Keyboard users encountered different barriers. Guided dialogs and onboarding overlays introduced additional focus stops, moved focus unexpectedly, and sometimes made it difficult to return to the point where the user had been working.
The flow also gave users very little sense of progress.
The underlying issue was becoming clear: onboarding had accumulated more steps, more information, and more interface behavior than users needed before reaching the product.
The Roadblocks
The accessibility barriers were spread across the journey rather than concentrated in one broken component.
Too much information was required before users could begin
The onboarding process had gradually become a place to collect profile information, preferences, configuration choices, and other account data.
From a business perspective, every question had a reason for being there.
From the user's perspective, every additional question created another decision, another instruction to interpret, and another opportunity for validation to fail.
Analysis of the flow showed that only around half of the information being requested was actually necessary before activation. The rest could be collected later without preventing users from entering the product.
This became one of the clearest root causes behind the eight-step journey.
Users often learned requirements only after making a mistake
Several fields depended on rules that were not sufficiently clear before input.
Instead of helping users succeed the first time, the interface made them discover requirements through validation.
This contributed to the 43% of onboarding sessions that required at least one validation retry.
For assistive technology users, recovery became even more demanding when the error message, field label, and problematic input were not programmatically connected.
Dynamic validation changed the interface without always communicating the change
Some fields validated after focus moved away. Others updated while the user was entering information. Visual feedback made many of these changes apparent to sighted users, but the same state changes were not consistently communicated through assistive technology.
When several errors appeared together, users also lacked a reliable way to understand how many problems existed and move directly to them.
Error recovery had effectively become a second navigation task inside onboarding.
Progress existed visually, but did not provide enough orientation
A multi-step journey needs to communicate more than the current screen.
Users need to know where they are, what has already been completed, what remains, and whether progress can be saved.
Instead, users could reach the fourth or fifth screen without having a meaningful understanding of how much work remained.
That uncertainty added to the cognitive load of an already lengthy process.
Overlays introduced a second navigation model
Guided tours, tooltips, and modal dialogs had originally been introduced to make onboarding more helpful. In practice, some of them increased complexity.
Users had to understand both the onboarding task itself and the mechanics of navigating the interface that was explaining that task. For keyboard and screen-reader users in particular, that meant unexpected focus movement, additional controls, and loss of context.
Why It Mattered
The business had already invested in making the primary product accessible. But the journey to setup the account - registration, preference selection, guided tours, and initial configuration - needed to be accessible as well. The accessibility of the destination has limited value otherwise.
Every user who abandoned onboarding represented an account that had already been acquired but had not reached activation. Marketing and acquisition had already done their job.
The loss was happening between acquisition and first meaningful product use. With 26% of users leaving before activation, the accessibility problem had become a measurable product and business problem rather than simply a compliance concern.
How We Approached It
We reviewed onboarding as one end-to-end task rather than auditing each screen independently. The complete journey was mapped from account creation through first product access, including every form state, validation event, modal, overlay, confirmation message, and transition.
Three areas became the priority.
First, we separated information required for access from information that could be collected later.
Questions that did not need to prevent entry into the product were moved out of initial onboarding. This allowed the core journey to be reduced from eight steps to four.
Second, we reviewed form comprehension and recovery.
Instructions, requirements, validation timing, error association, focus movement, and status announcements were considered together. The objective was to only get enough information from the user to succeed before submission and a clear way to recover when something went wrong.
Third, we improved orientation throughout the journey.
Progress indicators, page titles, step labels, modal behavior, focus management, and transition messaging were reviewed based on whether users could maintain context from one stage to the next.
This exposed problems that would have been easy to miss in a component-by-component accessibility review.
What Changed
The onboarding flow was reduced from eight required stages to four core steps. Non-essential profile questions and configuration choices were moved until after activation, allowing users to reach the product before being asked for information that was not necessary for account access.
Instructions were shortened and moved closer to the fields or decisions they related to. Password requirements and formatting expectations became available before submission rather than appearing only after validation failed.
Validation behavior was standardized. Error messages explained both what had gone wrong and what users needed to do next. Errors were associated with the relevant fields, and users were given a clearer path back to inputs requiring attention.
Dynamic status changes were reviewed so important feedback did not rely entirely on visual presentation. Progress communication was redesigned so users could understand their current step, the total length of onboarding, and what would happen next.
Modal dialogs and guided overlays were also simplified. Where an overlay created more complexity than value, it was removed instead of adding more accessibility behavior around it.
The Outcome
The impact became clearer when the same measures used to identify the problem were reviewed after the changes.
Faster onboarding
Reducing the initial journey from eight steps to four and moving non-essential questions until after activation reduced median onboarding time by approximately 27%, from around 11 minutes to 8 minutes. Users were reaching the product sooner without removing information the business still needed to collect.
Higher completion
The percentage of users successfully reaching activation improved by approximately 32%. The improvement reflected the combined effect of fewer required decisions, clearer expectations, better progress communication, and more predictable interaction behavior.
Fewer validation retries
Sessions requiring users to correct and resubmit information decreased by approximately 22% after requirements were surfaced earlier and validation behavior was standardized. This reduced both task time and the amount of error-recovery work users had to perform.
Better independent completion with assistive technology
Independent completion among keyboard and assistive technology test participants improved by approximately 45%. Improved focus management, clearer error association, reduced overlay complexity, and meaningful progress information all contributed to users being able to complete onboarding without assistance.
Stronger activation value
With more acquired users successfully progressing into the product, revenue associated with newly activated accounts also increased, therefore affecting more than WCAG conformance. They reduced friction at a point in the journey directly connected to activation and commercial value.
What We Learned
The most important finding was that the accessibility of the core application was only one part of the experience.
Users do not encounter products according to internal product architecture. They encounter a sequence of tasks. If onboarding prevents someone from reaching the main experience, an accessible core product has limited value.
The project also reinforced that onboarding accessibility is not solved simply by reducing the number of screens.
An eight-step flow can still work well if every step is necessary, understandable, predictable, and easy to navigate.
In this case, however, reducing eight steps to four addressed a measurable root cause because much of the information being collected did not need to block access in the first place.
The biggest improvement came from treating accessibility, usability, and product performance as parts of the same problem.
Fewer unnecessary decisions reduced cognitive load. Clearer validation reduced recovery time. Better focus management improved independent completion. And shortening the path to product value improved activation.
Accessibility was not an additional layer applied after onboarding was designed. It became part of deciding what onboarding actually needed to ask users to do.
.png&w=3840&q=75)
(1).png&w=3840&q=75)
.png&w=3840&q=75)