SEO & Search

Keyboard Navigation on a B2B Website: Review Menus, Dialogs, and the Route Back

A service page can explain the work clearly while its menu keeps a buyer outside.

Start at the homepage. Put the pointer aside. Find the service, open its information dialog, read the conditions, close it, and continue reading. Record where that journey stops.

This review follows one fictional purchase task. It gives a content owner evidence to hand to an accessibility specialist. It does not certify a website or supply code fixes.

The worksheet, role assignments, and examples below are proposed working methods. No observation describes this portfolio or an actual buyer.

Set the Journey Before Touching the Keyboard

The fictional company offers a documentation review. A buyer wants to find the service and determine which source files the review accepts. The homepage contains a Services control. Its expanded navigation links to a service-detail page. That page contains an “Accepted source files” information dialog with a close control. It contains no lead form.

Write that task at the top of the review record. Success means reaching the service, reading the file conditions, dismissing the dialog, and continuing to the service exclusions. Opening a menu alone does not complete it.

For a real session, record the browser and full version, operating system, keyboard settings, viewport, page address, and review date. State whether assistive technology is running. Record any setting that changes which controls receive keyboard focus.

For this invented example, assume a desktop Chromium browser, a physical keyboard, and system settings that permit keyboard navigation through links and controls. No screen reader is running. The exact browser version would belong in an actual observation. These assumptions describe the example, not universal browser behavior.

Begin from a freshly loaded homepage. Keep the pointer unused while completing the task. If a step requires a pointer to recover, record that recovery separately. It cannot stand in for a keyboard exit.

Keep the test narrow enough to repeat. A different navigation component, embedded application, or authenticated area needs its own journey and conditions.

Use Three Criteria as Separate Questions

WCAG 2.2 Success Criterion 2.1.1, Keyboard, addresses functionality available through a keyboard interface without prescribed keystroke timing. Its exception concerns an underlying function that depends on the movement path, rather than just the endpoints. W3C: Understanding Keyboard.

For this journey, ask whether the buyer can reach the service and open its information. Neither task requires freehand movement.

Success Criterion 2.4.7, Focus Visible, requires a mode with a visible keyboard focus indicator. W3C: Understanding Focus Visible.

Ask where the next keypress will act. A control might operate while its location remains visually unclear. Record operation and focus visibility separately.

Success Criterion 2.1.2, No Keyboard Trap, addresses leaving a component using a keyboard. An unusual exit method must be explained. W3C: Understanding No Keyboard Trap.

Ask whether the buyer can leave each component and resume the task. A menu that opens successfully can still interrupt the route back.

These three questions organize observations. They do not cover every accessibility requirement, assistive technology, or user need. A specialist determines the applicable findings and further review.

Follow the Component's Interaction Model

Do not write one universal sequence for every menu, disclosure, dialog, or custom widget. W3C notes that conventions vary with platforms and user agents. Keyboard guidance.

For each component, record the interaction model the implementation owner says it uses. Then record what the reviewer discovers without private instructions. A hidden technique known only to the developer needs attention even if someone can demonstrate it.

Use Tab and reverse navigation to explore the ordinary focus sequence in the stated environment. Record the actual activation keys used. Where a component uses arrow navigation or another interaction pattern, document that pattern separately.

Avoid treating a site's marketing label as its technical definition. A control called “menu” may expand ordinary links. Another may implement a widget with different keyboard behavior. Ask the specialist to assess the implemented structure.

For the fictional Services control, the proposed behavior is a disclosure exposing service links. The reviewer records entry onto the control, activation, movement through the revealed links, and departure. Those expectations belong to this example.

Do not require every paragraph to receive Tab focus. Reading information and moving between interactive controls are different tasks. This keyboard-only observation also leaves screen-reader reading behavior untested.

Copy a Component Worksheet

Use one record per component and state. Repeat the record when the same control behaves differently after navigation or dismissal.

Component and location: Name the visible control, page, and starting state. Include whether a menu or dialog is already open.

Expected task: Describe what the buyer needs to accomplish. Keep this independent of the reviewer's preferred key sequence.

Entry: Record the previous focus location and the key sequence that reaches the component. Note any skipped or unreachable control.

Visible focus: Describe the indicator and whether it remains identifiable during the attempt. Record an uncertain location as uncertain.

Activation: Record the key used and the resulting state. Distinguish no response, the wrong action, and the intended action.

Forward and reverse movement: Record the next and previous reachable controls. State whether the buyer can revisit an earlier choice.

Close or departure: Record the available keyboard exit, any visible instructions, and the result. Use “not tested” for an untried method.

Return location: After dismissal, record where focus lands and what the buyer can do next. Compare it with the agreed task sequence.

Evidence and ownership: Attach reproduction steps, observed impact, reviewer, environment, and the person responsible for specialist review.

Keep screenshots alongside the steps when useful. A screenshot can show an indicator, but it cannot establish which sequence opened a component or whether an exit worked.

Walk Through the Fictional Menu

Suppose the reviewer reaches Services and can identify its focus outline. Activating it reveals links. The reviewer reaches Documentation Review, follows it, and arrives at the service page. Reverse movement through the expanded navigation also works.

The worksheet records those actions as observed within the invented scenario. It does not infer behavior for other menus or devices.

Now suppose a second attempt reveals a different issue. The Services control operates, but its focus indicator disappears against the open panel. The reviewer can still count keystrokes and reach the service link.

Record the visibility problem without describing it as an unreachable service. Counting from a remembered starting point is the reviewer's workaround. It does not make the focused location visible.

A useful finding states the task and state: “With Services expanded, the reviewer cannot visually identify focus on the Documentation Review link.” Add the exact steps and environment before assigning it for review.

Compare that with “navigation is broken.” The broad description leaves the implementer guessing which state failed and what evidence would establish a correction.

Keep unrelated lead-form labels and validation outside this record. If the journey later reaches an inquiry form, create a separate review rather than claiming this menu check covers it.

Separate a Modal Boundary From an Inescapable Trap

A modal dialog can intentionally contain focus while it is open. W3C distinguishes that boundary from a trap when a keyboard user can dismiss the component and leave. No Keyboard Trap guidance.

For the fictional file-information dialog, the proposed sequence moves focus into the opened dialog. The buyer reads its conditions, reaches its close control, dismisses it, and resumes from the opener. Record those proposed expectations before comparing observations.

Test the implemented close control. Record Escape separately if the selected interaction model provides it. Do not turn one familiar key into a universal requirement for every component.

Suppose forward navigation cycles through the dialog's controls. That observation alone does not establish a trap. The reviewer still needs to examine whether an available keyboard action dismisses it.

Suppose instead that the close control cannot receive focus, the documented exit does nothing, and navigation continues cycling. Record each attempted route. The finding now describes a buyer unable to leave using the methods examined.

Return focus also deserves a field. If dismissal sends focus to an unexpected location, describe that location and the extra navigation required. Do not automatically assign every return-location defect to the No Keyboard Trap criterion.

Ask the specialist to distinguish the criteria and verify the underlying behavior. Keep the observation precise enough to support that judgment.

Hand Over a Reproducible Finding

This is a synthetic observation log for the fictional dialog. It demonstrates reporting structure and makes no claim about a live implementation.

Expected task: Read accepted file types, close the information, and continue to exclusions.

Starting state: Fresh service page. Dialog closed. Tab reaches “Accepted source files” through the page's normal sequence. In this synthetic implementation, the dialog contains a “Format details” link and a Close button. Its visible instruction says “Press Escape to close.”

Reproduction: Press Enter on the opener. Focus moves to “Format details.” Press Tab twice, then Shift+Tab twice. Record the focused element after each press. Press Escape, following the dialog's instruction. These keys describe this fictional component, not every dialog.

Actual result: Each Tab and Shift+Tab press returns focus to “Format details,” skipping Close. Escape leaves the dialog open. The buyer cannot resume the service page through the examined sequence.

User impact: The file information remains open and blocks continuation to exclusions. Recovery in this synthetic attempt requires abandoning the page or using the pointer.

Owner: Website implementation owner for reproduction. Accessibility specialist for criteria assessment. Content owner for maintaining the task and visible instructions.

Status: Open observation. No fix or retest is claimed.

Preserve failed attempts and successful alternatives in the same record. If an accessible close route exists, update the actual result. Do not leave an absolute “cannot exit” claim after discovering contrary evidence.

Keep comparison-table relationships in their own review. A working dialog says nothing about whether a table's values retain their headers.

Retest the Route, Then Bound the Conclusion

Agree on acceptance evidence before the owner changes the component. For this example, repeat the same journey and confirm a discoverable keyboard exit and a usable continuation point.

Record the changed version and environment. Preserve the original finding. Attach the retest result rather than replacing the earlier failure with an unexplained pass.

Complete the entire route after a local correction. An opener and close control might work while the next required service link remains unreachable. The buyer's task supplies the stopping condition.

Keep untested behavior visible: other browsers, responsive navigation, screen-reader interaction, additional dialogs, and authenticated content. A narrow review supports a narrow conclusion.

Use the findings within a broader B2B conversion review without attaching a conversion forecast. This worksheet measures observed task completion in specified conditions. It establishes neither improved rankings nor complete WCAG or legal conformance.

The route back belongs in the review.

Sources

Topic map: Accessible B2B websites

Articles in this cluster, then the 3 clusters that share the most topics with it.

This cluster: Accessible B2B websites

Neighbor: Technical SEO, crawling, and migrations

Neighbor: Customer and usability research

Neighbor: Topical authority and keyword strategy