Choose a comparable business question
Define the service, location and customer need before collecting answers. In a fictional example, a business compares suppliers for a specific regional maintenance job. Comparing that question with a broad national supplier list would answer a different question. Keep the wording and intended audience recorded. Avoid adding favourable details about your own business to one version while omitting similar context from the competitor comparison. The test should describe a realistic customer need rather than a prompt designed to produce a preferred winner.
Record the conditions and complete answer
Note the date, tool and visible settings or context relevant to the test. Preserve enough of the answer to distinguish a passing mention from a recommendation and any qualification attached to it. Record when the answer gives no named supplier. Do not present an unavailable result as a negative mention. Where conditions differ between observations, label the difference instead of combining everything into one apparently comparable total. The record should let a reviewer understand what was asked and what was actually returned.
Define what counts before counting
Decide how to record a brand mention, a linked source and a description of suitability. These are separate features of an answer. A competitor can be named as an example or excluded from a shortlist, so a name count alone can mislead. In the fictional maintenance comparison, record whether the answer discusses the required location or a different service. Keep ambiguous cases visible for review. Do not label every link a favourable citation without inspecting the sentence and destination it supports.
Inspect the cited material directly
Where an answer supplies a source, open the relevant page and check what it says about the business. Record missing, inaccessible or unrelated material. A source link does not automatically prove that every associated claim is supported. Compare your own published service details with the actual customer question as a content review, not an explanation of the platform’s internal selection process. Avoid claiming that one directory entry caused a mention simply because the directory appeared among the visible sources.
Repeat without overstating the pattern
Use a defined set of questions and a repeat schedule suitable for the review. Report the number of observations and the variation between them. A handful of answers supports a limited description of those tests, not a market-wide rank. Do not hide observations where your business disappears or the competitor set changes. Keep each comparison tied to its question and conditions. If the test design changes, start a clearly labelled series rather than treating the new results as directly equivalent to the old ones.
Turn observations into checkable content decisions
Identify practical gaps such as unclear service coverage or unsupported business claims and verify them against the company’s actual offer. Improve those pages because the information needs correction, without promising that the change will trigger a recommendation. Keep the change record separate from later mention observations. The comparison is useful when it identifies questions to investigate and evidence to improve. It does not reveal proprietary ranking rules or guarantee that a business will appear in another person’s answer.