TL;DR
- Gemini and Google AI Overviews are built by the same company but cite almost entirely different domains for the same prompts.
- AI Overviews behaves like a UGC-first surface — YouTube, Reddit, and forum content show up constantly. Gemini behaves like an authority-first surface, leaning on established publishers, reference sites, and Google's own properties.
- AIO citation is increasingly decoupled from classic organic rankings — ranking #1 doesn't guarantee the AIO citation, and vice versa.
- AIO's citation pool at the domain level is highly concentrated: a small number of sites absorb a disproportionate share of citations across unrelated queries.
- Gemini re-cites a narrow set of domains repeatedly rather than sampling broadly, which changes the optimisation math for anyone chasing visibility there.
Gemini vs AI Overviews (AIO) is a useful comparison for one simple reason: it isolates the surface from the company. Both run on Google infrastructure and both draw on Google's index, yet they pull from noticeably different citation pools when answering the same question. That gap tells you more about how each product is engineered — and who it's optimising for — than any amount of generic "optimise for AI search" advice.
Why two Google products cite different sources
AI Overviews is embedded directly in search results, generated from the same ranking and retrieval pipeline that produces the ten blue links. It is, structurally, closer to a search feature than a standalone assistant — the citations tend to track content that already performs well for the underlying query, including video, community threads, and UGC platforms. Gemini, by contrast, is a conversational assistant with its own grounding and retrieval behavior. When Gemini cites sources in chat, it draws from a retrieval layer that appears tuned toward recognized authority and structured reference content rather than whatever is currently ranking in search.
The practical result: ask the same question in Google Search (triggering an AI Overview) and in Gemini, and you'll often get two different citation lists with little overlap. This mirrors a pattern seen elsewhere in the ChatGPT vs AI Overviews citation overlap comparison — different AI surfaces, even from different vendors, rarely converge on the same source set. What's notable here is that this divergence holds even within a single company's own product line.
YouTube's outsized role in AI Overviews, not Gemini
One of the clearest asymmetries is video. AI Overviews frequently surfaces YouTube as a cited source, especially for how-to, product comparison, and troubleshooting queries — unsurprising given AIO's proximity to Search, where video results already rank prominently. Gemini's citation behavior for the same prompt types leans away from YouTube and toward text-based authority sources: documentation, established media, Wikipedia-style reference content, and in some cases other Google properties.
This isn't a minor stylistic difference — it changes what "optimising for Google's AI" even means. A brand investing in video content to capture AI Overviews citations may see little return if the audience question also surfaces in Gemini, where video carries far less weight. Teams tracking visibility across both surfaces need to treat them as separate optimisation targets, not a single "Google AI" channel. The comparison work in Which AI search engine your buyers actually use, by intent is a useful companion here — intent and surface both matter.
Gemini concentrates citations; it doesn't explore
A second pattern worth flagging: Gemini tends to re-cite the same handful of domains across many related prompts rather than widening its source pool. This looks less like broad web retrieval and more like a smaller, curated set of trusted domains being reused as grounding across a topic cluster. For brands outside that trusted set, breaking in is harder — not because content quality is lacking, but because Gemini's retrieval appears to favor domains it has already "seen" as authoritative, similar to how E-E-A-T signals function in traditional search (see why E-E-A-T matters more for AI search than traditional SEO).
AI Overviews, meanwhile, shows extreme concentration at the domain level too, but the composition of that concentrated set skews toward UGC and video platforms rather than publisher authority. Two different concentration patterns, two different barriers to entry.
AIO citation is decoupling from rankings — plan accordingly
A trend worth tracking independently of Gemini: AI Overviews citations are increasingly disconnected from a page's organic ranking position. A page ranking on page one may not appear in the AIO citation list for the exact same query, while a lower-ranked but differently structured page does. This means classic rank-tracking is no longer a reliable proxy for AIO visibility, and teams need direct citation monitoring rather than inferring AI visibility from SERP position. For practical detection, see how to check if ChatGPT is citing your website — the same logging and query-sampling logic applies to AIO and Gemini, just against different source pools.
Which surface to optimise for first
Given limited time, prioritise based on where your buyer actually asks the question and what content format you can realistically produce. If your audience researches inside Google Search (triggering AIO) and your content gaps are in video or community discussion, invest there first — UGC and video carry more weight. If your audience uses Gemini directly as an assistant, or your strength is in authoritative long-form and reference content, that's the better initial bet. Either way, avoid treating "Google's AI" as one undifferentiated target — run a free scan to see which URLs are currently surfacing across different assistants for your core questions, then decide where the gap is widest.
FAQ
Do Gemini and AI Overviews use the same index?
They both draw on Google's underlying web index and infrastructure, but their citation behavior differs — AI Overviews tracks closely with active search rankings including video and UGC, while Gemini's retrieval favors a narrower set of established, re-used domains.
Should I optimise content differently for Gemini vs AI Overviews?
Yes. AI Overviews rewards UGC, video, and content that already performs well organically, while Gemini tends to ground answers in authority-style reference content, so the content format and trust signals that work for one surface won't automatically transfer to the other.
Is ranking #1 in Google enough to get cited in AI Overviews?
No — AIO citation is increasingly decoupled from organic rank, meaning a page can rank well without being cited, and a lower-ranked page can still appear, so direct citation tracking is more reliable than inferring visibility from SERP position.
Sources
Drafted with AI assistance and published by CiteFlow. Sources are linked in the text.



