What does share of citation change for a founder who has been reporting share of voice?

I reported share of voice for years because it was the number a clip report could produce without arguing about what the clip did. It still answers that narrower question. It stops being the scoreboard when the buyer's first read is an answer engine. An engine does not award the loudest company. It retrieves sources. If your coverage is not among those sources, the impression count did not do the job you thought you bought.

Share of citation is the question I ask instead, on the prompts that match the buyer's wording. Was a credible source about the company cited? Not mentioned in a detector's window. Cited, with a page an engineer of the answer could point to. The formula and the tracking method live on the AuthorityTech page. The definition lives in the Machine Relations glossary. I am not republishing either. I am saying what changes in the founder's meeting when the scoreboard changes.

The meeting gets harder, and it gets shorter. You stop debating whose impression multiplier was fair. You open the answer and read the sources. If a community site leads the segment, the renewal question is being answered outside the outlets on the press list. That is not a reason to abandon earned coverage. It is a reason to ask which earned sources the engines in that segment actually retrieve, and to fund those, rather than the logo that looks best in a board slide.

I want the release id next to any share I quote. The Index is a window, not a law. A founder who cites 38 of 137 without the release will be quoting a number that the next release can move, and will have no way to say what changed. The reversal condition is the next release, read the same way: same segment, same question shape, new counts. If the sources move, the plan moves. If they do not, the plan does not get a new name to hide the miss.

Share of voice can stay in the appendix. It is a record of activity. Share of citation is a record of retrieval. I will not let a team average them into one index. Averaging a loudness measure with a retrieval measure produces a number that cannot tell you which problem you have. When the clips are plentiful and the answers are empty, you have a retrieval problem. When the answers cite you and the clips are thin, you have a coverage problem. They ask for different work.

The page I send an operator to is the metric page, not this one. The page I send a reader who wants the term is the glossary. The two sibling answers on this site are the budget test and the placement of GEO and AEO. Together they are the judgment. They are not the dataset.

The definition is share of citation in the Machine Relations glossary. The execution page is AuthorityTech's citation share definition, formula, and tracking method.