[]
en

Language

Book a call
Blog
Engines5 min read · By the New Map team · 23 October 2026

How to appear in Gemini recommendations?

Which sources the engines cited when asked how to appear in Gemini recommendations (7 answers)
Which sources the engines cited when asked how to appear in Gemini recommendations (7 answers)Bar chart, Which sources the engines cited when asked how to appear in Gemini recommendations (7 answers): the highest is support.google.com at 3; the 5 values range from 1 to 3. support.google.com was cited in 3 of 7 answers, arxiv.org and google.com in 2 each.support.google.com was cited in 3 of 7answers, arxiv.org and google.com in 2each.support.goo…3arxiv.org2google.com2ai-semantic…1altovista.co1
Source: New Map measurement, 19 October 2026 · How we measure

To appear in Gemini recommendations, state clear prices, drop hype and earn mentions on outside sources. DEJAN AI1 tested 14 October 2026 and found that pages featuring clear prices succeeded in 99.7% of their trials on Gemini 4 Argon. Start by reading which sources Gemini cites for your buyers' questions.

What do the engines answer when asked how to appear in Gemini recommendations?

Gemini publishes no recommendation checklist in these answers, so the first job is to read what the engines say. New Map's measurement2 counted 7 answers to this exact question across 3 engines over 2 weekly readings: Perplexity gave 2, Gemini 3 and ChatGPT 2. The engines named none of the tracked companies, and New Map was named in 0 of the 7 answers.

The cited sources tell more than the names. support.google.com appeared in 3 of the 7 answers, while arxiv.org and google.com each appeared in 2. Six other domains, among them cogny.com and altovista.co, appeared once each.

Our read: on this question the engines lean on Google's own help pages and research papers more than on any company. Seven answers is too few to call a trend. It shows where to start reading.

Which sources the engines cited when asked how to appear in Gemini recommendations (7 answers)
Which sources the engines cited when asked how to appear in Gemini recommendations (7 answers)Bar chart, Which sources the engines cited when asked how to appear in Gemini recommendations (7 answers): the highest is support.google.com at 3; the 5 values range from 1 to 3. support.google.com was cited in 3 of 7 answers, arxiv.org and google.com in 2 each.support.google.com was cited in 3 of 7answers, arxiv.org and google.com in 2each.support.goo…3arxiv.org2google.com2ai-semantic…1altovista.co1
Source: New Map measurement, 19 October 2026 · How we measure

How many sources does Gemini cite before it recommends anyone?

New Map's measurement asked 68 discovery questions about AI visibility tools and read 146 answers (how New Map measures). That is one week's reading of 68 questions about AI visibility tools on 3 engines (146 answers), not a trend: Perplexity cited a source in only 7 of its 68 answers, and Claude Code's 10 answers are too few to read alone.

New Map measured Gemini: 68 of 68 answers cited at least one source. That run counted 629 sources, which the same New Map measurement puts at 9.3 per answer. It named at least one brand in 33 of 68. Perplexity cited 0.3 sources per answer and Claude Code 7.3.

Across all 146 answers, reddit.com was cited in 20 (13.7%) and youtube.com in 16 (11%). wikipedia.org appeared in 1 of 146 (0.7%) (New Map measurement). These are tool questions only, so they are our reading of that set, not a Gemini norm. The chart shows Gemini with the highest figure of the three.

Sources Gemini cites per answer, against two other engines (146 answers)
Sources Gemini cites per answer, against two other engines (146 answers)Bar chart, Sources Gemini cites per answer, against two other engines (146 answers): the highest is Gemini at 9.3; the 3 values range from 0.3 to 9.3. Gemini cites 9.3 sources per answer, Claude Code 7.3, Perplexity 0.3.Gemini cites 9.3 sources per answer,Claude Code 7.3, Perplexity 0.3.Perplexity0.3Claude Code7.3Gemini9.3
Source: New Map measurement, 19 October 2026 · How we measure

How do you write pages that Gemini recommends?

Gemini 4 Argon is one version of the model, so these findings are version-specific. DEJAN AI (Dan Petrovic) published its head-to-head page tests. Four steps follow from them.

  1. State the price in plain text. On Argon, pages with clear prices won 99.7% of their tests, DEJAN AI's best feature result.
  2. Cut the hype. DEJAN AI reported that exclamation marks and all-caps words each lost all 150 of their tests on Argon.
  3. Do not count on tables. DEJAN AI measured table format winning 93.6% on 3.8 Flash and 31.5% on Argon.
  4. Do not lean on credentials. DEJAN AI reported that credential edits won only 15.4% across 2,249 live client experiments.

DEJAN AI's validation stage used 11 losing-page pairs per feature, so single results are early. The chart sets the price result beside the table result.

How pages win on Gemini: clear prices against table format
How pages win on Gemini: clear prices against table formatBar chart, How pages win on Gemini: clear prices against table format: the highest is Clear prices, Argon at 99.7%; the 3 values range from 31.5% to 99.7%. Clear prices won 99.7% on Argon; table format fell from 93.6% to 31.5%.Clear prices won 99.7% on Argon; tableformat fell from 93.6% to 31.5%.Clearprices,…99.7%Tableformat,…93.6%Tableformat,…31.5%
Source: DEJAN AI (Dan Petrovic), 14 October 2026

Which outside sources help a brand appear in Gemini recommendations?

The data shows these figures come from other engines or mixed sets. Our read is that outside mentions are worth pursuing, but no study here measured Gemini alone.

Kevin Indig3, 4 reported that AirOps found third-party signals drive 85% of brand visibility in language models. On 12 October 2026 he relayed an Ahrefs finding that YouTube mentions and branded web mentions correlated 0.7 with AI visibility in Google AI Overviews, against 0.3 for Domain Rating.

Paid publishers held 36.6% of citations in the Otterly5, 6 study published on 14 October 2026. Brand-owned pages held 38.9% in that Otterly study. The base was 450,473 citations (Otterly). Editorial sources held 11.7%. Your own site matters, and so do pages you do not control.

Indig also reported that 91% of citations appeared in only one of ChatGPT, Perplexity or AI Overviews. Gemini was not in that comparison, so plan sources engine by engine, as Generative Engine Optimization work does.

Outside sources: who owns the pages engines cite (450,473 citations)
Outside sources: who owns the pages engines cite (450,473 citations)Bar chart, Outside sources: who owns the pages engines cite (450,473 citations): the highest is Brand-owned at 38.9%; the 5 values range from 3.9% to 38.9%. Brand-owned pages hold 38.9% of citations, paid publishers 36.6%, editorial 11.7%.Brand-owned pages hold 38.9% ofcitations, paid publishers 36.6%,editorial 11.7%.Brand-owned38.9%Paidpublishers36.6%Editorial11.7%Community8.6%Institution…3.9%
Source: Otterly, 14 October 2026

Does the buyer's location change which brands Gemini recommends?

Chen and Koudas7 of the University of Toronto published a measured study. The network's country, not the declared location, determined Gemini's sources, which changed even when questions bore no geographic dependence.

The businesses the answers recommended changed along with the sources. The study ran ChatGPT from five network locations and with 19 declared locations, alongside Gemini from five network locations, executing each version twice to measure variability.

Otterly's study of 3.95M citations across seven engines, published 5 October 2026, found 64.9% of German AI citations carried a German-language or German-domain signal. That study was not Gemini-only.

Measuring this is hard. Semrush8 noted in October 2026 that Google Search Console's generative AI report shows impressions only, with no prompt-to-click attribution.

Gemini recommendations by location: how the network country reaches the recommended brands
Gemini recommendations by location: how the network country reaches the recommended brandsDiagram, Gemini recommendations by location: how the network country reaches the recommended brands: Network country, then Gemini's sources, then Recommended businesses. In the study, Gemini's sources and its recommended businesses changed with the network's country.In the study, Gemini's sources and itsrecommended businesses changed with thenetwork's country.Network countryGemini follows it, not…Gemini's sourcesMost change with countryRecommended businessesChange with the sources
Source: Chen and Koudas, University of Toronto, 16 October 2026

What to do

  1. Ask Gemini your category's buying questions from the country your buyers sit in, and record which domains it cites for each. Start with the three most cited and check whether you appear.
  2. Rewrite your pricing and product pages to state a clear price and remove exclamation marks and all-caps words, following DEJAN AI's 14 October 2026 Argon tests. Do not rely on tables.
  3. Earn mentions on the third-party and video sources Gemini cites in your own readings. New Map's plans and pricing cover measuring those answers and placing a brand in press and creator sources.

Frequently asked

Does Gemini cite sources in every answer?

In New Map's measurement, Gemini cited at least one source in 68 of 68 answers, 9.3 per answer. It named at least one brand in 33 of 68. That is one week's reading of tool questions, not a trend.

Do schema markup or llms.txt help a brand appear in Gemini?

Ahrefs found in May 2026 that 1,885 pages adding schema gained nothing against matched controls, with AI Overviews at -4.6%. That test covered content markup on already-cited pages, not Gemini itself, so treat schema as hygiene.

Can Google Search Console show Gemini recommendations?

Semrush reported in October 2026 that Google Search Console's Generative AI performance report displays impressions alone. It gives no prompt-to-click attribution and no queries, clicks or CTR data, so recommendations must be measured by asking the engines.

Sources

  1. DEJAN AI (Dan Petrovic), Gemini 4 Argon's personality: no hype, hard facts, fewer tables, dejan.ai, 14 October 2026
  2. New Map measurement, How we measure, 19 October 2026
  3. Kevin Indig, Shopping is part of the product, growth-memo.com, 12 October 2026
  4. Kevin Indig, How to plan 2027 when traffic is down and AI won't attribute, growth-memo.com, 5 October 2026
  5. Otterly, What AI Engines Actually Cite: 617,985 Citations Analyzed, otterly.ai, 14 October 2026
  6. Otterly, AI Search in Germany Plays by Different Rules. Here Is the Data., otterly.ai, 5 October 2026
  7. Chen and Koudas (University of Toronto), Location, Location, Generation: Where You Ask Matters for Generative Engine Optimization, arxiv.org, 16 October 2026
  8. Semrush, How to build an AI visibility dashboard with Semrush, semrush.com, 12 October 2026
  9. Seer Interactive, Competitive Analysis for Paid Media: How To Find the Broken Stage, seerinteractive.com, 7 October 2026

See where you stand

Book a call and find out how visible your brand is across the AI engines your customers already ask.

Book a call

For agencies: See the agency package