GEO Research Dashboard

Total Citations

3476

Explicitly attributed

Total Additional

7326

Shortlisted but not cited

Total Unsurfaced

1127

In SERP but not cited

TOTAL CONSIDERED

11929

Sum of all unique links

Unique Domains

1335

Across all runs

Total Coverage (Cited)

81.0%

Found in any SERP

Missing (Cited)

19.0%

Not in either

🏢 GPT Business

Cited

1637

Additional

2820

Unsurfaced

631

TOTAL CONSIDERED: 5088

Bing Cited Overlap

81.3%

Bing Add. Overlap

86.3%

Google Cited Overlap (control)

27.8%

Google Add. Overlap (control)

20.7%

Query 1 Overlap

63.7%

Query 2 Overlap

64.2%

UNSURFACED OVERLAP: Bing 9.7% | Google 3.6% (control)

Runs: 237 | Bing Results: 86938

👤 GPT Plus

Cited

1839

Additional

4506

Unsurfaced

496

TOTAL CONSIDERED: 6841

Bing Cited Overlap

67.6%

Bing Add. Overlap

56.3%

Google Cited Overlap

64.6%

Google Add. Overlap

52.1%

Bing Query 1 Overlap

54.1%

Bing Query 2 Overlap

52.4%

Google Query 1 Overlap

50.4%

Google Query 2 Overlap

46.4%

Total Coverage (Bing+Google)

80.6%

Google-only Coverage

13.0%

Missing (not in either)

19.4%

Query 1 Overlap

54.1%

Query 2 Overlap

52.4%

UNSURFACED OVERLAP: Bing 5.6% | Google 0.0%

Runs: 237 | Bing Results: 83984

Fan-Out Localization Patterns (GPT)

Three observed patterns where fan-out queries contain non-English language signals. 62 of 474 runs (13.1%) show localization signals.

Case 1: Foreign Prompt

Prompt is non-English; system adds English fanout

30

6.3% of runs

Case 2: Anomalies (Context-Triggered)

English prompt mentions foreign lang/region

9

1.9% of runs

Case 3: Anomalies (Untriggered)

English prompt, no foreign context → unexpected fanout

23

4.9% of runs

Case 1: Foreign-Language Prompts (30 runs) ▸

Run	Account	Prompt
P019_r1	enterprise	Quel est le meilleur logiciel de traduction instantané ?
P019_r2	enterprise	Quel est le meilleur logiciel de traduction instantané ?
P019_r3	enterprise	Quel est le meilleur logiciel de traduction instantané ?
P032_r1	enterprise	什么实时翻译软件好用？
P032_r2	enterprise	什么实时翻译软件好用？
P032_r3	enterprise	什么实时翻译软件好用？
P038_r1	enterprise	Anlık ses çevirisi yapan programlar nelerdir?
P038_r2	enterprise	Anlık ses çevirisi yapan programlar nelerdir?
P038_r3	enterprise	Anlık ses çevirisi yapan programlar nelerdir?
P055_r1	enterprise	Welche Transkriptionssoftware ist die beste?
P055_r2	enterprise	Welche Transkriptionssoftware ist die beste?
P055_r3	enterprise	Welche Transkriptionssoftware ist die beste?
P060_r1	enterprise	Anlık çeviri yapan programlar nelerdir?
P060_r2	enterprise	Anlık çeviri yapan programlar nelerdir?
P060_r3	enterprise	Anlık çeviri yapan programlar nelerdir?
P019_r1_personal	personal	Quel est le meilleur logiciel de traduction instantané ?
P019_r2_personal	personal	Quel est le meilleur logiciel de traduction instantané ?
P019_r3_personal	personal	Quel est le meilleur logiciel de traduction instantané ?
P032_r1_personal	personal	什么实时翻译软件好用？
P032_r2_personal	personal	什么实时翻译软件好用？
P032_r3_personal	personal	什么实时翻译软件好用？
P038_r1_personal	personal	Anlık ses çevirisi yapan programlar nelerdir?
P038_r2_personal	personal	Anlık ses çevirisi yapan programlar nelerdir?
P038_r3_personal	personal	Anlık ses çevirisi yapan programlar nelerdir?
P055_r1_personal	personal	Welche Transkriptionssoftware ist die beste?
P055_r2_personal	personal	Welche Transkriptionssoftware ist die beste?
P055_r3_personal	personal	Welche Transkriptionssoftware ist die beste?
P060_r1_personal	personal	Anlık çeviri yapan programlar nelerdir?
P060_r2_personal	personal	Anlık çeviri yapan programlar nelerdir?
P060_r3_personal	personal	Anlık çeviri yapan programlar nelerdir?

Case 2: Anomalies (Context-Triggered) (9 runs) ▸

Run	Account	Prompt	Non-English Query
P050_r1	enterprise	Can you recommend a free app for live time translation during a call with French speakers?	application gratuite traduction temps réel appel téléphonique traduire conversat
P058_r1	enterprise	What is the best software for Parsian/Farsi audio-to-text transcription?	بهترین نرم افزار تبدیل گفتار فارسی به متن Persian audio transcription software
P058_r2	enterprise	What is the best software for Parsian/Farsi audio-to-text transcription?	بهترین نرم افزار تبدیل گفتار فارسی به متن Persian audio to text transcription to
P058_r3	enterprise	What is the best software for Parsian/Farsi audio-to-text transcription?	نرم افزار تشخیص گفتار فارسی تبدیل صدا به متن بهترین
P047_r1_personal	personal	I need a Chrome extension for live Chinese to English translation that works better than the built-in live caption featu	Chrome 插件实时中文英语翻译 Chrome extension live Chinese English translate
P050_r2_personal	personal	Can you recommend a free app for live time translation during a call with French speakers?	application gratuite traduction en temps réel appels téléphoniques français engl
P058_r1_personal	personal	What is the best software for Parsian/Farsi audio-to-text transcription?	بهترین نرم افزار تبدیل صوت فارسی به متن audio to text Farsi transcription softwa
P058_r2_personal	personal	What is the best software for Parsian/Farsi audio-to-text transcription?	بهترین نرم افزار تبدیل گفتار فارسی به متن audio to text Farsi transcription
P058_r3_personal	personal	What is the best software for Parsian/Farsi audio-to-text transcription?	بهترین نرم افزار تبدیل گفتار فارسی به متن Persian audio to text software

Case 3: Anomalies (Untriggered) (23 runs) ▸

Run	Account	Prompt	Non-English Query
P007_r3	enterprise	Is there a real time audio to text translation?	实时音频转文字翻译 real time audio to text translate
P016_r2	enterprise	I need live transcribing while using online streaming or watching videos.	实时视频转录直播字幕 software live transcription for streaming
P018_r1	enterprise	Can you suggest a free website or program that can translate my 20-minute video and add subtitles?	免费工具翻译视频字幕自动 translate video subtitle free online
P018_r2	enterprise	Can you suggest a free website or program that can translate my 20-minute video and add subtitles?	無料 video 翻訳字幕自動サービス free translate subtitles video
P018_r3	enterprise	Can you suggest a free website or program that can translate my 20-minute video and add subtitles?	免费翻译视频自动字幕在线 tool free translate video subtitles
P031_r1	enterprise	Are there currently any free audio to text transcription services?	無料オーディオテキスト自動文字起こし無料 transcription services
P034_r1	enterprise	Is there a Windows application for live translation during calls in Google Meet or other meetings?	实时翻译应用 Windows during video calls meet translation software Windows
P043_r1	enterprise	Can you recommend software for live call subtitles in the same language and a separate translation line?	リアルタイム字幕ソフト翻訳ライブコールサブタイトル software live call captions translations
P048_r1	enterprise	Can I add a live translation app to calls?	ライブ通話翻訳アプリ通話中翻訳 phone call live translation apps
P048_r2	enterprise	Can I add a live translation app to calls?	リアルタイム翻訳通話アプリ live translation during calls iPhone Android
P054_r3	enterprise	What is the best instant translator for conferences?	实时会议同声翻译设备软件 conference instant translator review
P063_r2	enterprise	What is the best machine translation tool for live translation that saves context?	实时翻译工具保持上下文 machine translation tool live translation context saving
P072_r2	enterprise	Can you recommend the best video translator extension for YouTube?	おすすめ YouTube video translator extension translate YouTube captions add automatic
P007_r2_personal	personal	Is there a real time audio to text translation?	实时音频转文字翻译服务 real time audio to text translation
P007_r3_personal	personal	Is there a real time audio to text translation?	实时音频到文本转换实时 transcription tools
P017_r3_personal	personal	Can you recommend a text to speech service that can generate such a conversation with multiple voices?	サービス複数音声会話生成テキスト音声
P018_r2_personal	personal	Can you suggest a free website or program that can translate my 20-minute video and add subtitles?	gratis herramientas para traducir video y añadir subtítulos online free video tr
P020_r3_personal	personal	Request a tool to translate foreign audio to English in real-time during video playback.	实时外国语音翻译视频播放工具 real-time audio translation subtitles
P044_r1_personal	personal	Are there live translation software options for video chats like Zoom or Teams?	实时翻译软件视频聊天 Zoom Teams 翻译插件
P044_r3_personal	personal	Are there live translation software options for video chats like Zoom or Teams?	实时翻译软件视频会议 Zoom Teams live translate captions translation tools
P048_r1_personal	personal	Can I add a live translation app to calls?	¿puedo agregar traducción en vivo a llamadas telefónicas aplicación traducción e
P063_r1_personal	personal	What is the best machine translation tool for live translation that saves context?	实时翻译保存上下文机器翻译工具 live translation context aware
P063_r3_personal	personal	What is the best machine translation tool for live translation that saves context?	实时翻译工具保存上下文 best live translation tool that keeps context

GPT Business - Bing Page Distribution

Page	Matches
Page 1	1468
Page 2	333
Page 3	656
Page 4	692
Page 5	641
Page 6	560
Page 7	453
Page 8	437
Page 9	312
Page 10	271
Page 11	248
Page 12	202
Page 13	191
Page 14	166
Page 15	140
Page 16	103
Page 17	5

GPT Plus - Bing Page Distribution

Page	Matches
Page 1	521
Page 2	345
Page 3	663
Page 4	602
Page 5	667
Page 6	579
Page 7	590
Page 8	480
Page 9	433
Page 10	426
Page 11	391
Page 12	327
Page 13	293
Page 14	286
Page 15	264
Page 16	204
Page 17	6

GPT Business - Google Position Distribution

Bucket	Matches
organic_p1_1	88
organic_p1_2	59
organic_p1_3	34
organic_p1_4	37
organic_p1_5	41
organic_p1_6	39
organic_p1_7	33
organic_p1_8	26
organic_p1_9	21
organic_p1_10	14
organic_p2_1	17
organic_p2_2	18
organic_p2_3	14
organic_p2_4	18
organic_p2_5	16
organic_p2_6	18
organic_p2_7	9
organic_p2_8	9
organic_p2_9	17
organic_p2_10	8
organic_p3_1	11
organic_p3_2	7
organic_p3_3	4
organic_p3_4	7
organic_p3_5	6
organic_p3_6	6
organic_p3_7	8
organic_p3_8	3
paa1	9
paa2	7
paa3	6
paa4	1
video1	1
video2	2

GPT Plus - Google Position Distribution

Bucket	Matches
organic_p1_1	168
organic_p1_2	128
organic_p1_3	145
organic_p1_4	107
organic_p1_5	112
organic_p1_6	113
organic_p1_7	103
organic_p1_8	92
organic_p1_9	48
organic_p1_10	38
organic_p2_1	58
organic_p2_2	57
organic_p2_3	65
organic_p2_4	41
organic_p2_5	48
organic_p2_6	50
organic_p2_7	38
organic_p2_8	46
organic_p2_9	23
organic_p2_10	29
organic_p3_1	30
organic_p3_2	21
organic_p3_3	39
organic_p3_4	27
organic_p3_5	18
organic_p3_6	14
organic_p3_7	22
organic_p3_8	16
organic_p3_9	6
paa1	19
paa2	13
paa3	9
paa4	3
video1	1
video2	1
video3	1
discussion1	1
discussion2	1
discussion3	1

GPT Business - Top Invisible Domains

Domain	Count
en.wikipedia.org	116
arxiv.org	83
www.theverge.com	48
timesofindia.indiatimes.com	36
time.com	34
www.lifewire.com	32
www.wired.com	25
www.sfgate.com	24
maestra.ai	24
tomsguide.com	22

GPT Plus - Top Invisible Domains

Domain	Count
reddit.com	216
apps.apple.com	139
en.wikipedia.org	91
arxiv.org	70
transyncai.com	61
www.theverge.com	57
x-doc.ai	54
chromewebstore.google.com	54
jotme.io	44
www.wired.com	40

Enrichment Coverage & Label Distribution

Enriched URLs: 6837 / 6841 (99.9%)

Unlabeled URLs: 4

Structural Fidelity: 93.1%

Categorical Fidelity: 89.6%

Trend Correlation (r): 0.84

Filtered: Personal

Global Type

Type	Count	%
product_page	2410	35.2%
listicle	1999	29.2%
app_store_listing	645	9.4%
forum_ugc	451	6.6%
editorial_article	289	4.2%
other	281	4.1%
news_article	269	3.9%
reference	173	2.5%
documentation	143	2.1%
marketplace_directory	104	1.5%

Gemini Type

Business Type

Type	Count	%
product_page	959	52.1%
listicle	340	18.5%
app_store_listing	263	14.3%
reference	63	3.4%
other	47	2.6%
editorial_article	46	2.5%
documentation	41	2.2%
news_article	38	2.1%
forum_ugc	22	1.2%
marketplace_directory	16	0.9%

Selection & Attachment Drift Analysis

Percentage Point (PP) shift from search results to LLM selection. Positive = LLM "hunts" for this; Negative = LLM "avoids" this.

Drift types explained:

Selection Drift Feature lift when comparing cited URLs vs the full SERP menu (Menu → Order). Attachment Drift Feature difference between cited and additional (context-only) links.

Selection Drift (Menu → Cited, within SERP)

Showing thesis-grade Selection Drift (Weighted Average Lift). For Bing studies, ranks 6-10 may be omitted; weighted average is computed over ranks 1-5 only. GPT Plus vs Bing is computed only over ranks 1-5.

Listicles only

GPT Plus vs Google

Feature	Weighted Lift
has_tables	+12.23pp
is_current_year_2026	+4.49pp
has_numbered_lists	+2.20pp
has_bullet_points	+5.74pp
has_pros_cons	+3.38pp
has_clear_authorship	+1.83pp
expertise_signal_score	+2.20pp
freshness_cue_strength	+6.31pp
readability_score	-0.66pp
heading_density	+0.15pp

GPT Plus vs Bing

Feature	Weighted Lift
has_tables	+2.25pp
is_current_year_2026	+8.58pp
has_numbered_lists	-2.90pp
has_bullet_points	-8.82pp
has_pros_cons	-9.17pp
has_clear_authorship	+2.20pp
expertise_signal_score	-7.68pp
freshness_cue_strength	+11.97pp
readability_score	+0.80pp
heading_density	-0.40pp

GPT Business vs Bing

Feature	Weighted Lift
has_tables	+3.64pp
is_current_year_2026	-4.64pp
has_numbered_lists	+7.34pp
has_bullet_points	+3.29pp
has_pros_cons	-0.60pp
has_clear_authorship	-3.57pp
expertise_signal_score	-5.93pp
freshness_cue_strength	+2.15pp
readability_score	-0.47pp
heading_density	-0.78pp

Gemini vs Google

Feature	Weighted Lift
has_tables	+2.35pp
is_current_year_2026	-4.01pp
has_numbered_lists	+1.85pp
has_bullet_points	+1.74pp
has_pros_cons	+1.46pp
has_clear_authorship	+6.16pp
expertise_signal_score	-4.83pp
freshness_cue_strength	+1.44pp
readability_score	-0.34pp
heading_density	-1.94pp

Product pages only

GPT Plus vs Google

Feature	Weighted Lift
has_tables	+0.51pp
is_current_year_2026	+2.09pp
has_numbered_lists	-0.72pp
has_bullet_points	-0.06pp
has_pros_cons	+0.10pp
has_clear_authorship	+0.38pp
expertise_signal_score	-0.59pp
freshness_cue_strength	+0.67pp
readability_score	+1.22pp
heading_density	+3.11pp

GPT Plus vs Bing

Feature	Weighted Lift
has_tables	+4.93pp
is_current_year_2026	+5.24pp
has_numbered_lists	-0.22pp
has_bullet_points	+0.41pp
has_pros_cons	+0.37pp
has_clear_authorship	+5.32pp
expertise_signal_score	+1.80pp
freshness_cue_strength	+1.60pp
readability_score	+0.16pp
heading_density	+4.94pp

GPT Business vs Bing

Feature	Weighted Lift
has_tables	-0.37pp
is_current_year_2026	-0.30pp
has_numbered_lists	+0.84pp
has_bullet_points	-2.07pp
has_pros_cons	+0.61pp
has_clear_authorship	+0.39pp
expertise_signal_score	-1.09pp
freshness_cue_strength	-0.31pp
readability_score	-0.17pp
heading_density	+2.69pp

Gemini vs Google

Feature	Weighted Lift
has_tables	-5.11pp
is_current_year_2026	-0.58pp
has_numbered_lists	+8.73pp
has_bullet_points	-5.76pp
has_pros_cons	+0.26pp
has_clear_authorship	+1.03pp
expertise_signal_score	-10.34pp
freshness_cue_strength	-0.62pp
readability_score	+1.03pp
heading_density	-4.85pp

Statistical Significance (Two-Sample T-Test)

Comparing structural profile feature prevalence between models. p<0.05 indicates statistical significance.

GPT Business vs Plus

Feature	Ent %	Pers %	Diff	Sig
tables	21.3%	19.1%	+2.2%	p<0.01
numbered lists	50.3%	35.1%	+15.2%	p<0.01
bullet points	35.4%	29.0%	+6.4%	p<0.01
is current year 2026	15.1%	17.8%	-2.7%	p<0.01
clear authorship	33.6%	28.7%	+4.9%	p<0.01

Gemini Selection Preference (Order vs. Menu)

Feature	Order %	Menu %	Diff	Sig
bullet points	74.5%	78.7%	-4.2%	p<0.01
numbered lists	52.8%	48.9%	+3.9%	p<0.05
pros cons	38.1%	38.8%	-0.8%	ns
tables	33.4%	35.8%	-2.4%	ns
is current year 2026	17.0%	15.7%	+1.3%	ns
is vendor owned	23.1%	24.8%	-1.7%	ns

Attachment Drift (Cited vs Additional)

Showing Attachment Drift (Preference for citation from context).

Listicles only

GPT Plus

Feature	Drift
has_bullet_points	+5.64pp
has_clear_authorship	-5.86pp
has_numbered_lists	+5.79pp
has_pros_cons	+3.58pp
has_tables	+7.41pp
is_current_year_2026	+5.16pp
expertise_signal_score	+3.08pp
freshness_cue_strength	+3.57pp
heading_density	+3.89pp
readability_score	+0.36pp

GPT Business

Feature	Drift
has_bullet_points	-0.30pp
has_clear_authorship	-1.52pp
has_numbered_lists	+3.08pp
has_pros_cons	+3.31pp
has_tables	+5.63pp
is_current_year_2026	+7.83pp
expertise_signal_score	+8.05pp
freshness_cue_strength	+7.15pp
heading_density	-2.38pp
readability_score	-2.01pp

Product pages only

GPT Plus

Feature	Drift
has_bullet_points	+9.51pp
has_clear_authorship	-0.77pp
has_numbered_lists	+7.73pp
has_pros_cons	+0.23pp
has_tables	+3.70pp
is_current_year_2026	-3.61pp
expertise_signal_score	-1.59pp
freshness_cue_strength	-1.96pp
heading_density	-0.45pp
readability_score	+1.43pp

GPT Business

Feature	Drift
has_bullet_points	+5.85pp
has_clear_authorship	+0.10pp
has_numbered_lists	+5.12pp
has_pros_cons	+0.47pp
has_tables	+0.05pp
is_current_year_2026	-0.13pp
expertise_signal_score	+0.87pp
freshness_cue_strength	-2.03pp
heading_density	-0.85pp
readability_score	+2.11pp

Listicle Extraction & Bias Analysis

When an LLM cites a listicle (e.g. "10 Best AI Tools"), which items from that list does it pick? This section measures position bias, host exclusion, and reading accuracy.

Position Bias Within Listicles

A listicle ranks products #1, #2, #3 etc. Do models favour items near the top? Items analysed: GPT 596, Gemini 365.

Metric	GPT	Gemini	What it means
Median Picked Rank	3	3	Half of all picked items came from rank #3 or higher
Picked from #1 spot	23.3%	34.8%	How often the model chose the listicle's #1 item
Picked from top 3	54.2%	61.1%	% of picked items that were ranked #1-#3 on the listicle
Picked from top 5	77.7%	79.5%	% of picked items that were ranked #1-#5 on the listicle
Avg products per listicle	10.1	8.6	Average number of products on the listicle page

Takeaway: Both models strongly favour top-ranked items. Over half of all picks come from the top 3 — even though listicles average ~8-10 items. If you're not in the top 3 of a listicle, the LLM is much less likely to cite you.

Host Exclusion ("Does the listicle author get picked?")

Many listicles include the author's own product (e.g. HubSpot writing "10 Best CRMs" and including HubSpot CRM). How does the LLM handle this?

Metric	GPT	Gemini	What it means
Listicles where host has own product	431	405	Listicles where the site's own product appeared in the list
Host product skipped	63.6% (274/431)	71.9% (291/405)	LLM saw the host's product but chose NOT to recommend it
Host product picked	36.4%	28.1%	LLM chose the host's own product from the listicle
Random chance (1 / avg items)	10.7%	12.4%	If picking randomly, how often would the host be chosen
Bias multiplier	3.40×	2.27×	Host is picked this many times more often than random chance

Takeaway: The 2-3× bias multiplier looks striking, but vendors place their own product at rank #1 in over 93% of cases — and top-ranked items are already over-selected (position bias). After controlling for position, the residual host advantage shrinks to just +2.5–4.7pp, likely due to the more favorable language hosts use when describing their own products.

Does the LLM re-order items?

If a listicle says "#1 Notion, #2 Slack, #3 Zoom", does the LLM keep that order in its response?

Model	Kept Original Order
GPT	27.9% (166/596)
Gemini	29.5% (108/366)

~70% of the time, the LLM re-ranks the products — it doesn't just copy the listicle's order.

Reading Accuracy (Solo-Cited Listicles)

When the LLM cites only one listicle, how accurately does it extract product details from it?

Metric	GPT	Gemini
Accuracy score	4.71/5.0	4.81/5.0
Attribution failure	0.7% (5/689)	6.4% (27/421)

Reading is near-perfect. The real problem is attribution — sometimes the LLM links to a page that doesn't actually contain the product it described.