Skillux

FIELD GUIDE / Interaction quality

Loading, empty and error states: a practical UI checklist

An ideal-data screenshot cannot show whether an interface survives slow requests, missing records or failed submissions. Review the transitions between states before treating an AI-generated screen as ready for users.

By Skillux · Published · Updated · 4 min read

01

Model states around the task

Start with an action such as submitting an expense. List the conditions it can encounter: editable draft, validation failure, submission in progress, success and server failure. Add permission or connectivity conditions when they are relevant to the system.

Keep no-data states distinct. A first-time account with no expenses needs a starting action. A filter with no matches needs a way to adjust the query. A failed request needs recovery. Showing the same empty illustration for all three hides useful information.

02

Explain waiting without disrupting the user

Preserve useful content during a refresh where possible, and indicate which part is updating. Prevent accidental duplicate submissions while retaining a clear explanation of why the action is temporarily unavailable. Do not use an invented percentage for progress the application cannot measure.

03

Make errors identifiable and recoverable

Tell the person which input needs attention and preserve the rest of the form. Separate input mistakes from service failures: asking someone to change a valid email address will not repair an unavailable server.

Offer a realistic recovery path. Retry only when the operation can safely be repeated, and preserve the draft if the result is uncertain. For destructive or financial actions, coordinate the UI with backend protections against duplicate operations.

04

Exercise transitions, not just static screens

Use controlled responses or fixtures to reproduce each condition. Slow down the request, return an empty dataset, trigger a field error and simulate a service failure. Restore the normal response afterward and verify that retry or correction leads to the intended state.

Use a keyboard and inspect focus after each transition. A visible error message is not enough if the user cannot locate it or continue. A state checklist is a review aid, not a complete accessibility assessment.

  1. Open the initial state with both new and returning-user data.
  2. Submit invalid data and verify the field remains editable.
  3. Delay a valid request and attempt a repeated action.
  4. Fail the request, then recover without discarding the draft.
  5. Confirm success using the resulting record, not only a temporary toast.

Worked example

An expense submission state matrix

Illustrative specification; exact behavior depends on the application's data and transaction model.

Starting point

Form → success toast. Other conditions have no defined behavior.

More actionable

Draft: editable values. Invalid amount: text error associated with the field. Sending: visible pending state and duplicate-submit protection. Service failure: preserve draft and show a safe recovery action. Success: show the saved record and its reference. No expenses: explain how to create the first one.

This maps conditions to user-visible behavior. The implementation still needs to verify persistence, safe retries and assistive-technology announcements.

Take into your next review

Review checklist

  • First-use emptiness, no matches and failed loading are distinct.
  • Waiting states communicate what is happening.
  • Input errors include text and identify the relevant field.
  • Failures preserve recoverable user work.
  • Success can be verified beyond a transient notification.
  • Keyboard and status announcements have been reviewed.

Common questions

Should every loading message use an alert role?

No. Choose semantics appropriate to the urgency and context. Routine updates should not repeatedly interrupt the user. Test announcements with assistive technology rather than applying the same role everywhere.

Is a red border enough to show a form error?

No. W3C's explanation of error identification calls for a text description identifying the detected input error. Color can reinforce the message but should not carry it alone.

How do I stop a generated form clearing input after an error?

Keep the editable draft separate from the request result. Preserve field values on failure and show the relevant error. Clear the draft only after confirmed success or an explicit reset. Verify with a controlled failed response rather than a success-only demo.

What should a dashboard show when filters return no results?

Explain that the request succeeded but no records match the current filters. Keep active filters visible and provide a way to adjust or clear them. Distinguish this from an account with no records and from a failed request.

Sources & scope

Primary references checked on November 15, 2026. These sources support the principles cited beside them. The workflow, briefs and checklists are Skillux editorial recommendations; the examples are illustrative, with no claimed performance results.

W3C Understanding documents explain individual success criteria; this article is not a complete WCAG conformance assessment. References do not imply endorsement of Skillux. Read our editorial approach.