The Situation
One of our insurance clients had an online quote journey that ended with a plan comparison screen.
Three plans sat side by side - Basic, Standard, and Premium - with rows underneath for deductible, annual limit, roadside assistance, replacement vehicle, and other benefits. A highlighted ribbon marked the recommended plan, and each column ended with a Select button.
It was one of the most important screens in the quote journey and looked polished visually.
But the comparison had been built from div elements inside a CSS grid rather than a real data table. Green checks and gray dashes were also used to show whether each plan included a particular feature.
The client asked us to review the experience after accessibility testing showed that some users were struggling to compare plans and make a selection independently.
The Challenge
For a sighted mouse user, comparing the plans was straightforward. They could scan across a row and quickly see how each option differed.
For screen reader users, the same screen became a long sequence of disconnected information.
A value such as "$500" could be announced without making it clear which plan it belonged to or which feature it described. The check and dash icons did not communicate their meaning, and all three action buttons were announced simply as "Select."
Keyboard users could reach the buttons but had little context about which plan each button belonged to.
On smaller screens, the comparison also sat inside a horizontally scrollable container that was difficult to operate with a keyboard.
The visual relationships were clear.
The programmatic relationships were not.
What We Found
Our manual accessibility review identified several problems within the same component:
- No table semantics. Rows and columns existed only visually, so there was no programmatic relationship between a value, its plan, and its feature. This affected SC 1.3.1, Info and Relationships.
- Meaning relied on icons and color. Included and excluded states did not have sufficient text alternatives, while green and gray were also being used to distinguish them. This raised issues under SC 1.1.1, Non-text Content, and SC 1.4.1, Use of Color.
- Select buttons lacked context. Each plan ended with a button named only "Select," making the controls difficult to distinguish when encountered outside the visual layout.
- The scrollable area was difficult to operate by keyboard. Users could reach controls inside the comparison but could not reliably navigate the horizontal content.
- Sticky plan headers could obscure focus. On smaller screens, the fixed header could cover the control that currently had keyboard focus.
Individually, these appeared to be component-level accessibility issues.
Together, they meant some users could reach the end of the quote journey and still be unable to confidently compare or select a plan.
How We Addressed It
We worked with the client's product and engineering teams to preserve the visual comparison while rebuilding the structure underneath it.
The comparison was converted into a real table with a caption, column headers for each plan, and row headers for each feature.
For example:
<tr> <th scope="row">Roadside assistance</th> <td> <svg aria-hidden="true">...</svg> <span class="visually-hidden">Included</span> </td></tr>
The visual icons remained, but their meaning was also provided in text. Additional visual cues were introduced so availability did not rely on color alone.
Each Select button was given a distinct accessible name, such as "Select Standard plan," so users could understand the action even when navigating through a list of controls.
For smaller screens, the team also improved the responsive experience. Instead of forcing every user through a wide horizontal comparison, the plans could be presented in a stacked format while preserving the same information and relationships.
Keyboard focus behavior was reviewed alongside the sticky headers so the currently focused control remained visible.
Once the component was updated, we retested the complete comparison and selection journey using keyboard navigation and screen readers rather than relying on another automated scan.
That was important because the original component had looked complete visually while its underlying relationships remained inaccessible.
As we explain in our article on why a Lighthouse score is not a compliance result, automated testing alone cannot confirm that a complex interaction works for real users.
The Outcome
After remediation, the client saw a measurable improvement in users' ability to compare plans and make a selection independently.
In follow-up accessibility testing, successful independent plan comparison and selection increased from 68% to 80% among keyboard and screen reader users.
Accessibility-related support requests connected to the comparison step also decreased by 12% during the comparison period.
Users were better able to move through the table cell by cell, understand which feature and plan each value belonged to, distinguish between included and excluded benefits, and identify the correct Select button.
For the client, the improvements removed friction from one of the most important decision points in the quote journey without changing the visual comparison experience customers were already familiar with.
What We Learned
The problem was not that the comparison contained too much information.
The problem was that the relationships between that information existed only visually.
Once the plan names, features, values, states, and actions were connected programmatically, the same comparison became much easier to understand with a screen reader and keyboard.
This project also reinforced why accessibility needs to be considered at the component level.
A comparison table may appear on only one screen, but if that screen determines whether someone can understand their options and complete a purchase, its accessibility affects the entire journey.
For similar examples of complex visual interfaces, see our seat selection case study and live updates case study.
Next Step
Comparison, pricing, and quote screens carry important decisions, and they are often built from components whose visual structure is stronger than their underlying semantics.
If your product includes a comparison table or similar decision-making interface, book a call with an accessibility expert to review it against WCAG 2.2 AA.

.png&w=3840&q=75)
(1).png&w=3840&q=75)