// guide
Who writes the “best GEO agency” lists that Google’s AI Overviews cite? We checked 152 sources

Written by: Dmitry Filippov, Founder, GET-GEO.AI
Published: 2026-11-02 · Updated: 2026-11-05
// short answer
Mostly the agencies themselves. In 20 US searches on 2 November 2026, Google’s AI Overview cited 152 sources, and 79 were “best” lists that include the publisher’s own agency, 72 of them at number one. In 13 of the 20 searches, the first agency the overview named had ranked itself first in one of those cited lists.
- AI Overview appeared for all 20 “best GEO, AEO and LLM SEO agency” searches we ran from Dallas–Fort Worth on 2 November 2026, citing 152 sources (122 different pages).
- 79 of the 152 sources were lists in which the publisher’s own agency appears, and 72 of those put it first; 62 did not say so in the text.
- In 13 of 20 searches, the first agency named in the overview had ranked itself first in a list the overview cited; the top three never matched any list in order.
- YouTube was the largest single source with 39 citations (26%), mostly videos that rank the channel’s own agency; LinkedIn supplied 7, and Reddit 11.
- Since May 2026 Google’s spam policies define spam to include “attempting to manipulate generative AI responses in Google Search”.
What did we check?
We ran 20 English searches of the kind a buyer types when looking for a GEO agency, the work of getting a company named and cited in AI answers. Seven repeated, word for word, the titles of list articles we had seen ranking (“best AEO agencies 2026”, “top 10 GEO agencies for law firms 2026”); seven asked the same thing in other words; three were questions; three were control searches about neighbouring topics such as GEO for ecommerce and SaaS.
All searches ran on 2 November 2026 between 08:56 and 09:37 UTC in a signed-out incognito window, with personalisation off. Google’s footer placed every search in Dallas–Fort Worth, Texas. For each search we recorded the AI Overview text and the order of the agencies it named, every source in its source panel, and every list article in the top ten results. For each list we noted who published it, whether the publisher’s own agency was on it and where, and whether the page said so.
This is one snapshot per search. AI Overviews change between loads: our first search returned a different overview on each of three loads, and we kept the third. Our note on how we measure explains why we date every figure. We do not name the agencies or the authors in this article; the tables below use the figures only.
| Group | Searches | Example | Sources cited | Lists including the publisher |
|---|---|---|---|---|
| Exact list titles | 7 | best AEO agencies 2026 | 52 | 32 |
| Same need, other words | 7 | best generative engine optimization agency | 57 | 23 |
| Questions | 3 | What is the best GEO agency? | 18 | 8 |
| Control topics | 3 | best GEO agency for SaaS companies | 25 | 16 |
Who writes the lists AI Overview cites?
The agencies on the lists. Google showed an AI Overview for all 20 searches and cited 152 sources in total, 122 different pages from 71 domains. 102 of the 152 were “best” or “top” lists. In 79, the publisher’s own agency was on the list, and in 72 it was in first place.
The largest group, 36, were posts on agencies’ own blogs, where the domain makes the connection obvious. It is less obvious on other platforms. Of 27 YouTube citations of this kind, 19 never said that the channel belonged to an agency on the list; on LinkedIn, 4 of 6 articles did not mention it. The control searches looked no different: 16 of their 25 sources were lists that included the publisher.
A buyer reading these lists, or the AI Overview built on them, should assume that a list of agencies was written by an agency on it unless the page shows otherwise.
| Type of page | Cited | Lists including the publisher | Of those, no statement of the connection |
|---|---|---|---|
| Agency blog | 68 | 36 | 33 |
| YouTube video | 39 | 27 | 19 |
| Reddit thread | 11 | 1 | 0 |
| LinkedIn article or post | 7 | 6 | 4 |
| Media and directories | 5 | 0 | 0 |
| Other sites | 22 | 9 | 6 |
| Total | 152 | 79 | 62 |
Does ranking yourself first put you first in AI Overview?
In this niche, often. In 15 of the 20 searches, the first agency named in the overview was in first place on at least one list the overview cited. In 13 of those 15, that list was published by the same agency. Across all 79 citations of such lists, the publisher was named somewhere in the overview text 53 times.
The overview did not copy a list wholesale. It named 7.2 agencies on average, and its first three never matched the first three of any cited list in the same order. In our searches it looked more like the winner of one list followed by names gathered from several others.
Studies of software searches found such lists cited less often and rewarded less. Peec AI analysed 232,000 citations from software review prompts between December 2025 and February 2026: about 11% came from lists in which the publisher promoted itself, around 10% in Google AI Mode and Perplexity. In our agency searches the share was 79 of 152. Lily Ray checked 80 business software searches in AI Overviews between April and June 2026 and found that when a company’s own list was cited, the company was left out of the overview’s recommendations 69% of the time (224 of 323). In our data the agency behind a cited list was left out 26 times in 79.
A likely reason is that in this category the companies writing the lists are the same companies the buyer is searching for. We did not test that explanation; the figures describe 20 searches on one day.
Why is YouTube the largest single source?
Because agencies publish the same lists as videos. YouTube supplied 39 of the 152 sources (26%), 27 different videos cited in 15 of the 20 searches. No other domain came close: Reddit had 11 citations, the busiest single agency blog 9, LinkedIn 7.
33 of the 39 YouTube citations were videos ranking agencies, and in 27 the channel’s own agency was on the list. For these searches, a short video answering the exact question worked as a source in its own right: 18 channels supplied 39 citations, while the 68 citations of agency blogs were spread across 49 domains.
Does LinkedIn get into AI Overviews?
Occasionally, and mostly on long searches that match the article title. LinkedIn appeared in the overview’s sources 7 times, in 7 different searches: 4 of the 7 exact-title searches, 2 of 7 reworded ones, 1 of 3 questions and none of the controls. Six of the seven were list articles that put the author’s agency first; the seventh was a video post.
LinkedIn ranked better in the ordinary results than in the overview. It appeared in the top ten for 11 of the 20 searches, in first place for 3, yet the overview cited it in only 7. Ranking on the first page did not by itself make a page a source.
One author wrote four of these articles, one for each of four searches about agencies for law firms, each placing the same agency first without stating the connection, which we found only in public sources outside the articles. The overview cited all four articles but named that agency first in only one of the four searches.
Is ranking yourself first against Google’s rules?
Google has not said that self-promotional lists break its rules, but the rules now cover AI answers. In May 2026 Google added “attempting to manipulate generative AI responses in Google Search” to the definition of spam in its spam policies. In our reading, a list written mainly to be quoted by AI Overview, and not to help a reader choose, fits that description.
Google has not confirmed that it demotes these lists, and the published evidence is mixed. Ahrefs saw no clear ranking or citation penalty in December 2025, and Peec AI found no sign of AI platforms correcting for them through February 2026. In February 2026 Lily Ray reported that software and B2B sites publishing between 10 and 340 such lists had lost 29% to 49% of their Google visibility since mid-January; she described the lists as a pattern she saw, without proof that they caused the drops, and Google made no statement. The overviews we recorded in November 2026 still drew heavily on these lists.
What should an agency or B2B company do with this?
Write comparisons that help someone choose, and say whose they are. A list that includes your company can be useful if it states the criteria, gives each entry the same facts, carries a date, and says plainly that you are on it. In our data such statements were rare, which makes them a way to stand apart rather than a cost.
Use the buyer’s exact question as the title. On LinkedIn the articles that reached the overview matched long searches word for word. A short video answering the same question gives Google a second format to cite.
Before paying for a place in someone else’s list, check that AI answers cite that page: our guide on paid listicle placements shows how. If you are the buyer, read “best agency” lists as marketing, and use questions to ask a GEO agency to compare the agencies on them. If you want to know which pages AI answers cite when buyers ask about your category, our free audit runs ten buyer questions in one language and lists the sources behind each answer.
How do you check an agency named in an AI answer or a ranking?
Treat the name as a lead, not a verdict. Open the sources the answer cites and check who published each one. In our searches 79 of the 152 cited sources were lists written by an agency that included itself, so the first check is whether the page that put a name in the answer belongs to that same agency.
Then ask the same question again. AI answers change between runs: in our 33-answer ChatGPT test, around 40 providers appeared across 11 queries, and most were named once or twice. A name that comes back across repeats and across your own wording tells you more than one screenshot.
Finally, check the claim that got the agency recommended. If an answer says an agency works in nine languages or in your market, its own site should list those languages, say who does the work in each one and show a dated result you can verify. For multilingual or international GEO work, that list of languages and markets is often the reason the agency was named at all.
What are the limits of this study?
Twenty searches, one snapshot each, one morning, one US location, English only. Overviews vary between loads, so the counts describe what was shown on 2 November 2026, not a stable ranking. We matched agency names by hand and identified the connection between authors and agencies from the pages themselves and, for four LinkedIn articles, from public profiles. We will repeat the searches in February 2027 and publish the comparison.
Related questions
Is it wrong to include your own agency in a “best” list?
Not in itself. A comparison written by a company in the market can still help a reader if it explains the criteria, treats every entry with the same facts and says clearly that the author’s company is on it. The problem is a list that hides the connection: in our searches, 62 of 79 such lists did not mention it.
Will Google penalise self-promotional lists?
Google has not announced action against them. Since May 2026 its spam policies define spam to include attempts to manipulate generative AI responses in Search, which gives it grounds to act. On 2 November 2026 AI Overviews still cited such lists in all 20 searches we ran, so the risk is real but unmeasured.
Does a LinkedIn article work better than a blog post for AI Overviews?
Not in our data. LinkedIn supplied 7 of 152 sources, agency blogs 68 and YouTube 39. LinkedIn articles reached the overview mostly when their title matched a long search exactly, such as a niche plus “2026”. A blog post or video answering the same question had more chances.
Will these results still hold in three months?
Probably not exactly. Overviews changed between loads even on the same morning, and Google has signalled interest in curbing manipulation of AI answers. We will rerun the same 20 searches in February 2027 with the same method and publish what changed.
Related guides
Sources
- 01GET-GEO.AI research dataset: 20 Google searches from the US with AI Overview text, 152 overview sources and 117 list entries in the top-ten results, 2 November 2026
- 02Google Search Central — Spam policies for Google web search
- 03Google Search Central — Clarifying that spam policies apply to generative AI responses in Google Search (15 May 2026)
- 04Peec AI — Self-promotional listicles analysis: data from 232,000 citations (Tom Wells, 3 March 2026)
- 05Lily Ray — Why calling yourself the best could be helping your competitors win in AI search (17 June 2026)
- 06Search Engine Journal — Is Google finally cracking down on self-promotional listicles? (Lily Ray, 5 February 2026)
- 07Ahrefs — Do self-promotional “best” lists boost ChatGPT visibility? (Glen Allsopp, 4 December 2025)
- 08Our guide: should you pay to be listed in “best X” articles?
- 09Our guide: questions to ask a GEO agency
- 10Our note: how we measure
// share
How to cite this page
You may quote and reuse this content with attribution and a link to this page.
“Who writes the “best GEO agency” lists that Google’s AI Overviews cite? We checked 152 sources” — GET-GEO.AI, 2026-11-05. https://get-geo.ai/en/guides/best-agency-lists-in-ai-overviews