Published Growing Search case
AI share of voice for mapping software
Tie product capabilities to specific work questions and preserve the comparison set behind a visibility score.
Results reported by Growing Search
- Share of tracked brand mentions
- 71%
Brand plus seven competitors, across monitored questions
The question
Does the proposed comparison preserve the same question and competitor sets?
A comparison-cohort acceptance check
Decide whether the observation sets are comparable before interpreting a share. A changed comparison set requires its own record.
Freeze the question-set version
Keep the exact task wording with each observed answer. Record additions separately so the new set can be distinguished from the earlier cohort.
Freeze the compared-brand set
Retain the set used in the approved comparison. A change to that set needs a named revision rather than a silent update to the summary.
Retain the observation conditions
Save the assistant and available model identifier with the date and response. Keep missing conditions explicit in the working record.
Decide how to compare a changed cohort
If the question or compared-brand set changes, preserve both versions. Re-run the earlier cohort when a like-for-like comparison is needed.
| Published action | Recommended check | Record to retain |
|---|---|---|
| Use-case pages | Test the product task against the current interface before accepting its explanation. | Task input, expected output, page and product version. |
| Mention tracking | Preserve the compared set and question version with every summary. | Question revision, compared set, assistant, date and answer. |
| Change | Record | Proposed comparison decision |
|---|---|---|
| Question added | Old and new question-set versions. | Keep the earlier cohort available for re-observation. |
| Compared brand changed | Both compared sets. | Label the revised comparison separately. |
| Assistant or model changed | Available identifier and date. | Distinguish that condition in the report. |
| Product task changed | Product version and accepted task. | Recheck the page explanation before repeating the observation. |
Use-case pages in the published project
A mapping product's specific capabilities were missing from AI recommendations. Growing Search created question-specific use-case pages and tracked mentions against seven competitors. Its reported share uses mentions across monitored questions, without publishing the question list, model breakdown or observation period.
A product task with verifiable steps
Choose a software question that the product team can demonstrate. Identify the input the user starts with and the output the task should produce. A page explaining that sequence gives the reviewer something more precise to check than a general feature claim.
Ask the product owner to test the instructions against the current interface. Keep prerequisites and limitations near the steps they affect. A reader should be able to determine whether the product handles the task before starting a trial or requesting a demo.
- Use actual support or sales questions to select the task.
- Record the capability and product version supporting the explanation.
- Check the page after a feature or interface update.
The denominator behind share of voice
In a new measurement plan, name the competitors and freeze the question set for each comparison period. Record the assistant and observation date with every answer. Adding competitors or changing the questions can alter the denominator, so preserve those changes alongside the score.
The AI citations guide helps distinguish a mention from a linked source. Keep both separate from product sign-ups or accepted demos. A use-case page can be evaluated for accurate task completion while its search and commercial effects are measured independently.
Prepare a check for your own site
Choose the checks that fit your site and add the baseline you will use. The download includes this analysis, related guides and the original Growing Search publication.
Findings
Preserve the old comparison when the scope changes. A new result may be useful for the revised task set while remaining unsuitable for a direct comparison with the earlier summary. The cohort record makes that distinction reviewable.