Position-Adjusted Word Count is a visibility metric introduced in the 2024 Princeton GEO study to measure how prominently a source appears inside an AI-generated answer. Rather than counting whether a page was cited, it counts how much of the page's wording the answer actually used, then weights that by where and how prominently the citation appears, so being quoted at length and early scores higher than a passing link near the end.
It exists because a raw citation count is too blunt: two sources can both be "cited" while one supplies a single clause and the other supplies half the answer. Position-Adjusted Word Count separates them, which is why the study used it to rank content edits, adding statistics lifted it by about 41%, keyword stuffing lowered it.
Read it as a research measure, not a live dashboard number. It was computed on a simulated engine over a fixed benchmark, so it is directional evidence about what makes a source prominent, not a KPI any engine reports today. The practical equivalent you can track is your own citation coverage and share of voice per engine over time.