Design advisor rules and test recommendations

Start with constraints and preferences#

A hard constraint must be satisfied for a product to be appropriate. A preference helps order otherwise suitable products. For example, a required device connection is different from a colour preference. Do not make every answer a strict filter: a catalogue with incomplete attributes can otherwise produce no results.

Map each choice to usable data#

Inspect the actual products behind an answer choice. A question about keyboard layout needs a reliable layout attribute; a free-text description that only says ‘international’ may not establish the requested layout. Correct missing or inconsistent source information before treating it as a confirmed match.

Use a small test matrix#

Write down the answers you choose and the result you expect before testing. This makes it easier to spot whether a later rule change fixed one case while breaking another.

• Typical choice: several plausible products should remain.
• Narrow choice: only products with verified required attributes should remain.
• Conflicting choices: the flow should explain the limitation rather than invent a match.
• Incomplete catalogue data: inspect the missing attribute before relaxing a requirement.

Check the explanation as well as the ranking#

A product can be suitable while its generated explanation is inaccurate. Verify every claimed benefit against the source, and check links and product variants. Retest after editing questions, rules, or the catalogue; previous results belong to the earlier configuration.

Technical details and methodology#

For the collection steps, eligibility rules, calculation units, and limits behind this workflow, use the implementation-reviewed technical guides below.