When a listening query changes, which mention count is comparable?

A lighting brand tracks mentions with a saved query for “Harborlight” alongside “lamp” or “light.” Its team learns that customers also call one product “HL Mini” and adds that nickname to the query. The next weekly report shows more mentions. Did conversation grow, or did the query start finding posts it previously missed?

Keep the original query as version 1, including its keyword logic, English-language filter, exclusions, and reporting effective dates. Create version 2 with the nickname added; don’t overwrite version 1. For one week, run both against the same sources and time window. Save the version used with each result, then compare unique mention IDs: which mentions appear in both versions, and which appear only in version 2? Inspect a sample of the additions too. “HL Mini” may refer to something unrelated, so extra volume is not automatically useful volume.

That parallel week gives the report an explanation for the step change. Afterward, version 2 can become the reporting definition, while the version 1 record remains available to explain older counts. If language or exclusion rules change later, give those edits their own version as well. A small change log—query text, effective dates, reason for change, and the parallel comparison—is enough to make the trend auditable without building a larger reporting system.

What do you record when a saved listening query changes, and how do you show the resulting break in a mention-count chart?