Jakob Nielsen’s writing about accessibility repeatedly asks whether people can accomplish the work a website promises. That question gives technical evaluation a practical companion: observe the experience of people using the service with their usual tools. His Flash-era work is historical, but it provides a useful way to examine the connection between access, understandable interaction and successful task completion.
Read the Flash-era work in its context
In Making Flash Usable for Users With Disabilities, published October 13, 2002, Nielsen described early research involving four Flash applications. He discussed the unfamiliar interaction model, moving elements, missing explanations and the distance between related controls and feedback. He also identified the limited pool of suitable applications at that early stage.
The article documents questions raised as Flash MX introduced accessibility support. It should be read as a dated study of that environment, with its stated limits. Flash Player is now retired; a modern project needs current standards, supported technologies and evidence from its own users.
Connect access with a meaningful outcome
In his November 20, 2005 essay Accessibility Is Not Enough, Nielsen encouraged evaluating realistic work with people using their own assistive technology. The enduring question is whether a product enables the job, not merely whether a supplier says it has accessibility features. His suggested informal check belongs to that essay’s specific setting; it is not a universal sample size or conformance method.
For a current service, state the task in terms of the result: find the appointment location, correct a booking, compare two options or obtain a receipt. That helps a reviewer notice when all the controls are reachable but the information still leaves a person unable to decide what to do.
Plan an observation that respects the participant
Recruit participants whose experience and access methods are relevant to the service. Explain the session, obtain consent for any recording, agree suitable arrangements and compensation, and let people use familiar tools where practical. Prepare test accounts and fictional information so a session does not require exposing personal records.
W3C’s user-involvement guidance explains the value and limits of participation in evaluation. Plan coverage across different needs and combine observations with technical expertise. One person’s successful session provides evidence about that session; it leaves other users, states and environments to examine.
Give the goal without naming the control you expect someone to choose. Observe where they look, what they understand, what they try and whether help is needed. If you intervene to keep the session moving, record that intervention so the result remains interpretable.
Turn an observation into a design question
This example is an editorial illustration, not a result from Nielsen’s study. It shows how a task observation can identify a design relationship worth investigating. The repair still needs technical review, including how the update is exposed and announced, and a repeat of the task with relevant users.
Record the evidence without attributing the problem to a participant’s ability. “The confirmation was outside the viewed area and was missed” points to a testable interface question. “The user was confused” leaves the team with little guidance.
Combine methods and state the limits
Use standards evaluation to examine defined requirements, expert review to investigate patterns, and user sessions to understand actual task experience. Maintain separate records of each method’s scope and findings, then prioritize the changes together. The W3C evaluation overview is a useful starting point for that broader plan.
Nielsen’s historical contribution is most useful when it prompts better questions about a present design. Preserve dates and context when citing his work, and make contemporary decisions from current requirements and observed behavior. A clear result states which tasks improved, under what conditions and what remains to be checked.