Describe an input in ordinary language
Choose a fictional input with a small number of fields. Write what each field means and which values the learning exercise expects. Do not begin with a large copied customer file. A short sample lets you inspect every row and discuss how a rule should behave without exposing personal or commercial information.
Separate missing information from an actual negative answer. An empty field may mean unknown rather than no. If the rule treats those cases differently, write that down before implementation. Avoid silently filling gaps from assumptions about a person, town, language or business. A predictable decision table should make its uncertainty visible.
List the rule and the intended outcome
Create columns for condition, outcome and explanation. Use an outcome that matches the scope: show a note, place a fictional row in a review list or ask for clarification. Do not include a real external action in the first browser-local exercise. The explanation should tell the user why that outcome was selected.
In an illustrative Harare workshop, a sample request with a missing topic could be placed in needs-clarification, while a complete fictional request enters ready-for-human-review. These labels do not confirm a service or automatically contact anyone. They are examples of a transparent rule, not an application form or a live business process.
Look for overlap and gaps
Read each condition against the other conditions. Ask what happens when several appear to apply or none matches. Choose an explicit way to handle those cases and keep the explanation with the table. An unresolved conflict should not be hidden by whichever line of code happens to run first.
Ask another person to follow the table using a sample you did not use while writing it. Note where they need to guess. That hesitation can reveal an undefined term or a missing case. Correct the table before polishing the interface. A clearer label may solve the misunderstanding more effectively than adding another automatic step.
Preserve the rule version
Give the decision table a date and a simple version reference. Keep the matching sample data and expected outcomes with it. If the rule changes, update the explanation and tests together rather than leaving the interface to describe an earlier version. A result is easier to assess when its rule is identifiable.
JSON and browser text-output references explain possible implementation tools, but a valid data format cannot decide whether your rule is fair or useful. Those are design and organisational questions. This exercise helps expose them before real deployment; it does not replace proper authorisation, professional judgement or review of a higher-impact automated decision.
For one row, ask a second reader to invent an input that nearly fits. Discuss why it belongs elsewhere and add that distinction to the table rather than relying on memory.
Sources & further reading
Use these references to check the underlying guidance and any current local requirements. Practical worksheets and illustrative examples in this guide are original educational material.
This guide is general information. Source notes are not a claim of independent professional review.