A difficult comparison table does not condemn the whole catalogue

A buyer on a phone can see product values but loses the column headings while moving across a comparison table. The catalogue receives complaints, and a complete rebuild is proposed. Before accepting that conclusion, establish what the current system can preserve and what actually fails.

Reproduce the reading task

Select two fictional products whose differences matter: one has a specified material and another has a different operating condition. Ask the reviewer to identify which value belongs to which product and unit at the troublesome viewport size.

Record the browser, viewport and interaction used. A cropped screenshot can identify a visual symptom, but the implemented table must be inspected to understand headings, focus, scrolling and the information supplied to assistive technologies.

Separate content from presentation

If the underlying product data correctly records the values and units, a component repair may be possible. If the source stores a single unstructured text field for every variant, the problem may also involve content modelling and editing.

Inspect how staff updates the table. A repair that requires editors to maintain duplicate phone and desktop values can introduce a new consistency risk. Prefer a design whose presentations derive from the same approved facts where the platform supports it.

Compare two explicit scopes

A focused repair could improve the table's reading order and responsive behaviour, with checks for representative products and supported input methods. A wider replacement might be justified if the platform cannot express the required data relationships or is no longer supportable under the project's needs.

Each option should state what remains unresolved. A table correction is not a full accessibility audit, and a redesigned site does not guarantee search or sales outcomes. The decision should be proportionate to the evidence.

Include the cost of transition

A replacement must account for existing product links, documents, editing roles and late changes. A small repair needs its own release and recovery arrangement, but may avoid that broader move. Compare the entire task rather than just the first visible improvement.

Test the chosen result with the same comparison that exposed the problem, then add an absent value, a long product name and the relevant language edition. Confirm that the user can still identify the product, characteristic and unit.

Bring the current page to website improvement planning. The evidence can support a bounded correction or a well-founded redesign, without deciding the answer from the age of the interface alone.