Browse documentation

Workflows

Investigate a visibility change

Use this workflow when visibility, recommendation rate, sentiment, share of voice, or citations change materially. The objective is to explain the observed movement with answer-level evidence before assigning work.

A repeatable method for validating and explaining a movement in a Digraph metric.

1. Establish the comparison scope

Record the tracker, current and comparison dates, platforms, prompt count, region, language, and competitor set. A configuration change can produce a different aggregate without a meaningful change in AI responses.

2. Locate the contributing prompts

Use the affected platform and date to find prompts whose observed answer outcome changed. Check whether the run set is complete before interpreting a percentage or rate.

3. Read the underlying answers

Compare answer text, mentioned brands, recommendation classification, and citations. Separate observed facts from hypotheses about why the response changed.

4. Decide and record the next action

Create a task only after the evidence supports an action. Link the task to the prompt, answer, audit issue, or source context used to make the decision; otherwise record the item as an unresolved observation.

Was this page helpful?