Teams routinely track ChatGPT alone and treat it as a proxy for all AI search. When ChatGPT, Google AI Mode, Perplexity, Claude, and Gemini answer the exact same buyer questions with live search turned on, how often do they pick the same brands?
We measured 889 matched buyer questions across ChatGPT Search (gpt-6), Google AI Mode, and Perplexity (sonar) from November 1 through November 19, 2026; 189 matched buyer questions across all five live search engines from October 15 through November 19, 2026; and 6,730 Claude Opus 5 answers across 434 buyer questions from September 1 through October 16, 2026.
Across 889 three-engine buyer prompts, 189 five-engine prompts, and 6,730 Claude Opus 5 answers, 66.8% of brand-question appearances hit only one engine and no pair agreed on the #1 brand even 30% of the time.
Two-thirds of brand mentions appear on only one of the five live engines
Across 189 matched buyer questions answered by all five live search engines (ChatGPT Search, Google AI Mode, Perplexity, Claude, and Gemini) between October 15 and November 19, 2026, we recorded 4,786 unique brand-question pairs. Fully 66.8% (3,197 pairs) appeared on just one engine.
Cross-engine coverage collapses quickly as you add engines: 17.6% (844 pairs) appeared on two engines, 7.8% (371) on three, 4.3% (205) on four, and only 3.5% (169 pairs) appeared on all five engines answering the exact same prompt.
- Brand appearances across all 5 live engines
- 3.5%
169 of 4,786 pairs · 66.8% single-engine
189 matched buyer prompts across ChatGPT, Google, Perplexity, Claude, Gemini
- Highest pairwise #1 brand agreement
- 29.5%
ChatGPT vs Perplexity (889 matched prompts)
All 10 five-engine pairs landed between 20.8% and 28.6%
- Shorter answers on Perplexity vs ChatGPT
- 3.4x
2,269 vs 7,644 avg chars · 7.08 vs 7.34 brands
Same median 7-brand shortlist packed into a third of the length
- Decisive comparison rate on Claude Opus 5
- 88.2%
Across 19,259 facts · vs 75.7% on ChatGPT Search (14,561 facts)
9.90 avg brands and 63.9% repeat #1 retention across 6,730 answers
How many engines named the same brand on the same question (4,786 pairs)
- 1 engine only66.8%3,197 pairs
- 2 engines17.6%844 pairs
- 3 engines7.8%371 pairs
- 4 engines4.3%205 pairs
- All 5 engines3.5%169 pairs
Takeaway
No pair among the five live search engines reaches 30% agreement on #1
Even the two closest engines disagree on the #1 pick more than 7 times in 10. Across 889 matched buyer questions in November, ChatGPT Search (gpt-6) and Perplexity (sonar) agreed on the #1 brand just 29.5% of the time (across 814 prompts naming brands on both). ChatGPT Search and Google AI Mode agreed 28.4% (843 pairs), and Google AI Mode and Perplexity agreed 25.9% (829 pairs). Jaccard shortlist overlap across the trio ranged from just 0.218 to 0.256 (2.52 to 2.84 shared brands out of ~7.2 named).
Adding Claude and Gemini on the 189-prompt five-engine panel confirmed the ceiling: every single one of the 10 pairwise combinations landed between 20.8% (Claude vs Gemini) and 28.6% (Claude vs Perplexity).
Pairwise #1 brand agreement across engines
All 10 pairwise engine comparisons across ChatGPT Search, Google AI Mode, Perplexity, Claude, and Gemini.
Same #1 (%) · 10 results
- Claude vs Perplexity28.6%
- ChatGPT Search vs Perplexity28.2%
- Google AI Mode vs Perplexity27.6%
- Gemini vs Google AI Mode26.9%
- ChatGPT Search vs Google AI Mode25.3%
- Claude vs Google AI Mode25.3%
- Gemini vs Perplexity24.9%
- ChatGPT Search vs Gemini21.9%
- ChatGPT Search vs Claude21.2%
- Claude vs Gemini20.8%
Explore the data (10 rows)
| Claude vs Perplexity | 168 | 28.6 | 2.49 |
| ChatGPT Search vs Perplexity | 181 | 28.2 | 3.09 |
| Google AI Mode vs Perplexity | 185 | 27.6 | 2.85 |
| Gemini vs Google AI Mode | 182 | 26.9 | 2.93 |
| ChatGPT Search vs Google AI Mode | 182 | 25.3 | 2.76 |
| Claude vs Google AI Mode | 170 | 25.3 | 2.08 |
| Gemini vs Perplexity | 181 | 24.9 | 3.1 |
| ChatGPT Search vs Gemini | 178 | 21.9 | 2.63 |
| ChatGPT Search vs Claude | 165 | 21.2 | 2.08 |
| Claude vs Gemini | 168 | 20.8 | 2.59 |
Takeaway
Gemini surfaced 78% more distinct brands than ChatGPT Search on the same prompts
Across the 189 five-engine prompts, Gemini cast by far the widest brand net. It generated 2,201 total brand mentions covering 968 distinct brands, 78% more unique brands than ChatGPT Search's 543 distinct brands (1,431 total mentions). Google AI Mode placed second at 633 distinct brands (1,562 mentions), ahead of Perplexity at 590 (1,462) and Claude at 563 (1,195).
What didn't differ was how far down the domain authority curve each engine reached. Median domain authority rank stayed clustered between 27,952 (Claude) and 33,113 (Gemini), and 66.2% to 71.2% of brand mentions on every engine came from domains ranked outside the top 10,000.
Distinct brand breadth and authority reach across 5 engines (189 prompts)
Distinct brands · 5 results
- Gemini968
- Google AI Mode633
- Perplexity (sonar)590
- Claude563
- ChatGPT Search (gpt-6)543
Explore the data (5 rows)
| Gemini | 968 | 2,201 | 33,113 | 18.5 | 67.6 |
| Google AI Mode | 633 | 1,562 | 31,961 | 19.9 | 68.2 |
| Perplexity (sonar) | 590 | 1,462 | 31,585 | 18.8 | 68.9 |
| Claude | 563 | 1,195 | 27,952 | 19.2 | 66.2 |
| ChatGPT Search (gpt-6) | 543 | 1,431 | 31,585 | 16.3 | 71.2 |
Takeaway
Perplexity writes 3.4x shorter than ChatGPT while packing dense alternatives
Across 889 matched buyer questions in November, ChatGPT Search, Google AI Mode, and Perplexity named almost identical numbers of brands per answer: a median of 7 brands on all three engines, averaging 7.34 on ChatGPT, 7.35 on Google, and 7.08 on Perplexity. Perplexity actually topped both rivals in ultra-crowded answers naming 10 or more brands (26.1% vs 24.0% on ChatGPT and Google).
The huge gap was response padding. ChatGPT Search averaged 7,644 characters per answer and Google AI Mode averaged 3,462 characters, while Perplexity delivered its 7-brand shortlist in just 2,269 characters (3.4x shorter than ChatGPT).
Average answer length on 889 matched buyer questions
- ChatGPT Search (gpt-6)7,644 chars7.34 brands · 24.0% 10+
- Google AI Mode3,462 chars7.35 brands · 24.0% 10+
- Perplexity (sonar)2,269 chars7.08 brands · 26.1% 10+
Takeaway
Google AI Mode leaves fewest empty answers (1.3%) despite 4.6x fewer citations than Perplexity
Citation volume doesn't predict whether an engine actually names brands. On the 889-question trio, ChatGPT Search attached 10.60 citations per answer and Perplexity attached 9.97, yet Perplexity still returned zero brands on 6.1% of prompts and ChatGPT on 4.2%.
Google AI Mode attached only 2.18 citations per answer (roughly a fifth of ChatGPT and Perplexity) yet hit the highest brand-inclusion share at 98.7%, leaving just 1.3% of prompts with zero brands.
Average citations attached per answer (889 matched prompts)
- ChatGPT Search (gpt-6)10.60 cites95.8% named brands
- Perplexity (sonar)9.97 cites93.9% named brands
- Google AI Mode2.18 cites98.7% named brands
Takeaway
Perplexity leads in alternative framing; Google AI Mode leads in incidental drops
Evaluating 21,111 brand judgments across the 889 matched prompts under Parse's standardized brand-evaluation rubric shows distinct editorial styles. Google AI Mode had both the highest recommended share (65.8% across 7,066 judgments) and the highest incidental name-drop rate (19.5%), while keeping alternative framing lowest at 3.5%.
Perplexity kept incidental mentions lowest at 9.5% (across 6,835 judgments) and pushed alternative framing highest at 11.2%, slightly ahead of ChatGPT's 9.6% (across 7,210 judgments). Both ChatGPT (10.0%) and Perplexity (10.3%) also placed brands in direct head-to-head comparisons more often than Google AI Mode (6.5%), while Google carried the highest warning share (1.46% vs 1.20% on Perplexity and 0.83% on ChatGPT).
How 3 engines frame mentioned brands (% of brand judgments)
- ChatGPT Search
- Google AI Mode
- Perplexity
- Recommended
- ChatGPT Search58.8%
- Google AI Mode65.8%
- Perplexity61.9%
- Incidental
- ChatGPT Search13.7%
- Google AI Mode19.5%
- Perplexity9.5%
- Alternative
- ChatGPT Search9.6%
- Google AI Mode3.5%
- Perplexity11.2%
- Compared
- ChatGPT Search10.0%
- Google AI Mode6.5%
- Perplexity10.3%
Takeaway
Claude Opus 5 names 9.90 brands per answer and picks decisive winners 88.2% of the time
To gauge how Anthropic's flagship performs at larger longitudinal depth, we tracked 6,730 Claude Opus 5 answers across 434 buyer questions from September 1 through October 16, 2026. Claude Opus 5 averaged 9.90 brands per answer and kept the same #1 brand in 63.9% of 6,296 consecutive same-question repeats.
In direct comparisons (counting subject_better or subject_worse as a decisive verdict), Claude Opus 5 declared a clear winner 88.2% of the time across 19,259 extracted comparison facts (Sep 1 to Oct 16), compared with 75.7% across 14,561 ChatGPT Search comparison facts in the October 12 to November 19 GPT-6 era window.
- Average brands per Claude Opus 5 answer
- 9.90
Across 6,730 answers (434 buyer questions)
- Consecutive repeat #1 brand retention
- 63.9%
Across 6,296 consecutive prompt pairs
- Decisive head-to-head comparison verdicts
- 88.2%
19,259 Claude facts vs 75.7% across 14,561 ChatGPT Search facts
Takeaway
Full 889-prompt scorecard: ChatGPT Search, Google AI Mode, and Perplexity
Sort and compare every headline metric across the November 889-prompt panel below, from mention coverage and length to citation counts and brand-level framing shares.
889-prompt engine scorecard (November 1 to 19, 2026)
Avg chars · 3 results
- ChatGPT Search (gpt-6)7,644
- Google AI Mode3,462
- Perplexity (sonar)2,269
Explore the data (3 rows)
| ChatGPT Search (gpt-6) | 95.8 | 7.34 | 10.6 | 7,644 | 58.8 | 9.6 | 10 | 2.225 |
| Google AI Mode | 98.7 | 7.35 | 2.18 | 3,462 | 65.8 | 3.5 | 6.5 | 2.387 |
| Perplexity (sonar) | 93.9 | 7.08 | 9.97 | 2,269 | 61.9 | 11.2 | 10.3 | 2.162 |
Takeaway
What to do with these numbers
Break out AI visibility reporting by engine instead of averaging into a blended score. When 66.8% of brand mentions occur on only one engine and no engine pair hits 30% #1 agreement, an aggregate index conceals complete blind spots on Google AI Mode, Perplexity, Claude, or Gemini.
Match your content assets to each engine's format bias: tight, citation-backed alternative tables for Perplexity (2,269 chars, 11.2% alternative framing), structured fit tradeoffs for ChatGPT (7,644 chars, 10.60 cites), and decisive head-to-head comparison proof for Claude.
Takeaway
Study scope and counting rules
This study covers three matched panels. First, 889 matched buyer questions (857 organic) answered across ChatGPT Search (gpt-6), Google AI Mode, and Perplexity (sonar) from November 1 through November 19, 2026, generating 21,111 brand-level judgments under Parse's standardized brand-evaluation rubric (7,210 ChatGPT, 7,066 Google AI Mode, 6,835 Perplexity).
Second, 189 matched buyer questions answered across all five live search engines (ChatGPT Search, Google AI Mode, Perplexity, Claude, and Gemini) from October 15 through November 19, 2026, yielding 4,786 brand-question pairs. Third, 6,730 Claude Opus 5 answers across 434 buyer questions from September 1 through October 16, 2026 (6,296 consecutive repeat pairs and 19,259 comparison facts, compared against 14,561 ChatGPT Search comparison facts from October 12 through November 19, 2026 where decisive = subject_better or subject_worse).
Pairwise #1 brand agreement is computed strictly over pairs where both compared engines named at least one brand. Long-tail shares count brands whose domain authority rank exceeds 10,000.
How we measured
889 matched buyer questions (857 organic) answered across ChatGPT Search (gpt-6), Google AI Mode, and Perplexity (sonar) from November 1 through November 19, 2026 (21,111 brand judgments under Parse's standardized brand-evaluation rubric); 189 matched buyer questions across all five live search engines (adding Claude and Gemini) from October 15 through November 19, 2026 (4,786 brand-question pairs); plus 6,730 Claude Opus 5 answers across 434 buyer questions and 19,259 comparison facts from September 1 through October 16, 2026.
- brand appearances hitting all 5 live engines
- 3.5%
169 of 4,786 pairs (66.8% appeared on only 1 engine)
- pairwise #1 agreement across 5 engines
- 20.8% to 28.6%
No pair reached 30% on 189 five-engine or 889 trio prompts
- distinct brands surfaced by Gemini vs ChatGPT
- 968 vs 543
78% wider unique brand pool on Gemini (189 matched prompts)
- decisive comparison verdicts (subject better/worse)
- 88.2% vs 75.7%
Claude Opus 5 (19,259 facts) vs ChatGPT Search Oct 12-Nov 19 (14,561 facts)