Mention Position

AI answers to commercial questions usually take the form of a ranked or semi-ranked list: a lead recommendation, two or three alternatives, sometimes a budget pick. Users anchor on the first name, just as they once clicked the first result. Studies of recommendation lists consistently show steep attention decay after the top entries, so a brand mention in position one is worth several mentions in position six.

Position also encodes the model's confidence. Engines tend to lead with the brand most strongly associated with the query in their training data and retrieved sources, making average position a readable proxy for how firmly the category is linked to you.

For each collected answer, every named brand is extracted in order of first appearance: position 1, 2, 3, and so on. Aggregated over a prompt set, this yields average position per brand, per engine, per topic, plus distribution stats like the share of answers where you lead. Position data is most powerful when paired with mention rate: two brands can both appear in 60 percent of answers while one leads in most and the other trails in all.

Because answers vary between runs, position, like every AI metric, should be averaged over repeated daily samples rather than judged from single answers.

Climbing positions is about strengthening the brand-category association engines perceive. Leading brands tend to dominate the sources answers are grounded in: they top the cited review rankings, anchor the comparison tables, and appear first in listicles. Earning those leading slots in heavily cited third-party content, and publishing definitive own content for the query, pulls your average position up over time.

Geonimo records the position of every extracted mention alongside sentiment, so teams can distinguish prompts where they merely appear from prompts where the AI actually recommends them first.