Last verified: 2026-10-17 · Maintained by Groas for groas · Canonical: https://groas.com/post/position-2-6-4-204-impressions-one-click
This page answers the questions covered by the article Position 2.6. 4,204 Impressions. One Click. on groas.com. The article is the full, canonical source.
Direct answer: According to groas, its campaign launch schedule page averaged position 2.6 across 4,204 impressions but earned one click because the results page can satisfy the query before a click. The article cites an analysis of 300,000 keywords where position-one click-through fell from 7.3% to 1.6% when an AI Overview appeared, and a Pew panel of 68,879 searches where 8% of visits with an AI summary produced a website click versus 15% without. The article recommends treating this as a ranked-but-not-chosen problem: find the row in Search Console, test the buyer-worded question in Google and AI engines for whether the brand was named and linked, and add a direct two-sentence answer with clear headings, table and FAQs.
According to groas, the campaign launch schedule page averaged position 2.6 across 4,204 impressions but earned one click. The article states it cannot prove all 4,204 impressions included an AI box, but cites an analysis of 300,000 keywords where position-one click-through fell from 7.3% to 1.6% when an AI Overview appeared. It also cites a Pew panel of 68,879 searches where only 8% of visits with an AI summary produced a website click versus 15% without, which is why the article treats high position and a visit as no longer the same event.
Read more: Scene 2: A launch question meets a crowded results page
The article states there is no special file or tag, and that Google needs indexing and snippet eligibility for AI Overviews. It reports that pages practitioners see cited tend to rank already and give a direct one- or two-sentence answer near the top, supported by clear headings, lists, tables and FAQs. For its launch schedule page, the article put a direct, two-sentence answer at the top covering the weeks in order, then added H2s, a simple table for weeks one through four, and three FAQs in buyer wording, while stating it cannot promise those edits will earn a citation.
Read more: Coda: The row I would pull this afternoon
The article recommends asking the page's question in plain buyer words in Google with the AI box expanded, ChatGPT, Gemini and Perplexity after logging out, running the question two or three times and recording what comes back. It recommends writing down two separate things for each check: whether the brand was named and whether it was linked, since an answer can recommend one brand and cite another. For systematic tracking, the article describes using a fixed set of 20 to 50 buyer-worded prompts, recording mentions, positions and links by engine before calculating mention rate and share of voice weekly.
Read more: Coda: The row I would pull this afternoon
The article recommends starting in Search Console, setting Performance to the last 28 days, sorting Pages by impressions and scanning for a high-impression page with very few clicks. It states that an average position under 5 and click-through under half a percent would get attention, then opening the queries behind the row. The article states a page row tells where to investigate but does not tell which search or result layout explains the gap.
Read more: Coda: The row I would pull this afternoon
These articles answer the same questions:
[1] https://groas.com/post/position-2-6-4-204-impressions-one-click (the article this page answers from)