Lorixen.
EN

A Comparison Tool Needs Two Different Cells: Unknown Attribute or Unsupported Feature

A comparison tool on a business website usually ends up with cells that are not a plain yes. A visitor sees a dash and wonders: does this option lack the feature, or does the site simply not know? Those are different statements. Treating them the same misleads readers in one direction or the other. This article proposes a way to keep them distinct in the data and on the screen. All options and attributes below are invented for illustration, and nothing here makes claims about any real provider.

Three cell meanings, not two

For every attribute of every compared option, allow exactly one of these values:

  • Supported: the editor has a note that the option includes this.
  • Not supported: the editor has a note that the option does not include this.
  • Not yet checked: nobody has recorded an answer.

A further value, Depends on the plan or setup, is useful when the answer is conditional. Add it only if your table needs it, because every extra value must be explained to the reader.

The key rule: a blank is never a no. An empty field in the data must render as Not yet checked, not as Not supported.

Show the difference visibly

Use text, not only symbols or colour, so the table works for everyone. A proposed legend: Included, Not included, Not checked. Give Not checked a neutral style and a short line beneath the table: Cells marked Not checked mean we have no recorded answer. They do not mean the feature is missing.

If a cell has a recorded answer, add the date the editor last confirmed it, or a note field for what the answer was based on, such as public product information. Do not display a date you cannot back with a record.

Inputs for the example

Take a made-up comparison of three invoicing tools, Tool A, Tool B and Tool C, across attributes: recurring invoices, multi-currency, export to spreadsheet. Suppose the editor recorded these notes: Tool A recurring invoices included, Tool B multi-currency not included, Tool C export included. Everything else is unrecorded.

Decision branches

For each cell, ask in order:

  • Is there a recorded note saying included? Show Included.
  • Is there a recorded note saying not included? Show Not included.
  • Is there a recorded note saying conditional? Show Depends, with the condition in a note.
  • Otherwise show Not checked.

When filtering, for example show only options with multi-currency, decide what Not checked does. A cautious approach: exclude Not included, keep Not checked in a separate group labelled Not confirmed, and never merge it into the matching group.

Stop conditions

Do not publish the table if more than the editor is comfortable with sits at Not checked for an attribute that is central to the comparison. That threshold is your own choice. Do not publish if any Not included cell has no recorded note behind it, since an unsupported negative claim can be as misleading as an unsupported positive one.

Worked example

A visitor filters for export to spreadsheet. Tool C appears under Included. Tools A and B appear under a separate heading, Not confirmed, with the explanation that no answer is recorded for them. The visitor is not told these tools lack export, and not told they have it. The page suggests checking the provider's own current information before deciding.

Small acceptance test

Build the invented dataset above and check each of the following.

  1. Render the table. Pass if all unrecorded cells read Not checked and none read Not included.
  2. Edit one blank cell to Not included without a note. Pass if the publishing check rejects it.
  3. Apply the export filter. Pass if Tool C is listed as Included and the other two appear under Not confirmed.
  4. Read the table with a screen reader or text-only view. Pass if each cell's meaning is conveyed in words.
  5. Delete a recorded note. Pass if the cell reverts to Not checked rather than keeping its old value.

Conclusion

A comparison table is only as honest as its empty cells. Separate unknown from unsupported in the data, show the difference in words, handle filters carefully, and refuse to publish negatives without notes.

Sources

    WhatsAppLet’s talk