Parse
IndexPricing
Sign inCheck your brand
  1. Research
Research
Does AI praise brands even when it ranks them lower?

Does AI praise brands even when it ranks them lower?

Published Aug 12, 2026Updated Oct 12, 2026

On this page
  • The finding
  • How we measured
  • Sources
  • More like this
  • The finding
  • How we measured
  • Sources
  • More like this

Dimitry ApollonskyFounder, Parse

DataWhat reviewed means

Positive language can sound like an endorsement even when a brand is placed well down a recommendation list. We wanted to see how praise varies with recommendation position.

We compared description language with the rank of the associated brand in ChatGPT Search and Google AI Mode answers from May 24 to July 16, 2026. This keeps what AI says about a brand separate from where it places that brand among alternatives.

Praise does not mean first place. AI used positive language in 81.2% of descriptions of brands ranked fourth or lower in our ChatGPT Search and Google AI Mode sample.

Four in five descriptions at fourth or lower were positive

AI used positive language in 640,708 of 788,976 descriptions for brands ranked fourth or lower, or 81.2%. A description is an adjective, short phrase, or positioning claim attached to one ranked brand in one observed AI answer.

Positive language is not evidence that a brand won the recommendation. Use sentiment to understand framing, then use recommendation rank to understand the shortlist decision.

Takeaway

Treat positive sentiment as framing, not as proof that the brand won the recommendation.

First place was more positive, but praise extended down the list

Positive language appeared in 367,591 of 390,696 descriptions for first-ranked brands, or 94.1%. The rate was 88.7% for brands ranked second or third and 81.2% for brands ranked fourth or lower. First place exceeded fourth or lower by 12.9 percentage points.

Rank and sentiment move together, but they are not interchangeable. Benchmark a brand against others at similar positions before treating a favorable-sentiment score as a competitive lead.

Positive description rate by recommendation rank

  • First94.1%367,591 of 390,696
  • Second or third88.7%550,897 of 621,074
  • Fourth or lower81.2%640,708 of 788,976
  • 0%50%100%

ChatGPT Search had more than twice Google's rank gap

ChatGPT Search's positive-description rate fell from 91.2% at first place to 74.0% at fourth or lower, a 17.2-point gap. Google AI Mode fell from 96.4% to 89.4%, a 7.0-point gap. The ChatGPT Search gap was 2.5 times as large.

A blended sentiment benchmark hides how differently the engines separate winners from the rest of the list. Track the same priority prompts on both engines and compare sentiment among brands at the same rank within each one.

First-to-fourth positivity gap

  • ChatGPT Search17.2 points
  • Google AI Mode7.0 points
  • 05101520

Takeaway

Keep separate rank-and-sentiment baselines for ChatGPT Search and Google AI Mode.

Positioning claims lost the most positivity below first place

At fourth or lower, 289,109 of 323,173 adjectives were positive, or 89.5%. The rate was 77.0% for other phrases and 67.4% for positioning claims. Positioning claims fell 24.4 points from their 91.8% first-place rate.

A single overall sentiment score can hide which language changed. Separate simple adjectives from claims about who the brand is for, what it leads on, or where it fits.

Positive description rate at fourth or lower

  • Adjectives89.5%289,109 of 323,173
  • Other phrases77.0%303,237 of 394,035
  • Positioning claims67.4%48,362 of 71,768
  • 0%50%100%

Excellent was the most common positive word at every rank

The word excellent appeared 5,376 times at first place, 10,557 times at second or third, and 11,240 times at fourth or lower. Strong, recommended, best, and comprehensive also appeared in the leading terms for every rank group.

Generic praise is shared across the list. Audit the specific attributes and positioning claims attached to the brand instead of counting common approval words as differentiation.

Common positive words by recommendation rank

First · 5 results

  • Excellent5,376
  • Best4,972
  • Recommended3,922
  • Comprehensive3,671
  • Strong2,842
  • 02,0004,0006,000
Explore the data (5 rows)
Excellent5,37610,55711,240
Strong2,8426,8919,657
Recommended3,9224,7447,804
Comprehensive3,6715,0694,514
Best4,9725,1024,488

Lower-rank positivity ranged from 70.1% to 90.5% by industry

Among industries with at least 1,000 descriptions in each displayed rank group, Clothing and Apparel had 3,709 positive descriptions in 4,100 at fourth or lower, or 90.5%. Blockchain and Cryptocurrency had 1,598 in 2,278, or 70.1%. Its 23.5-point gap from first place was the largest in the displayed set.

Use an industry baseline before calling a brand's language unusually positive. The spread identifies where a rank-and-language review is useful and does not show that industry caused the difference.

Selected fourth-or-lower positive rates

  • Clothing and Apparel90.5%3,709 of 4,100
  • Manufacturing87.6%5,841 of 6,670
  • Food and Beverage85.8%5,994 of 6,985
  • Financial Services77.6%28,509 of 36,747
  • Biotechnology74.8%1,709 of 2,286
  • Blockchain and Cryptocurrency70.1%1,598 of 2,278
  • 0%50%100%

Alphabet and Microsoft drew the most praise at fourth or lower

Alphabet had 6,934 positive descriptions among 8,658 at fourth or lower, or 80.1%. Microsoft had 5,439 of 6,898, or 78.8%. Grafana, Notion, Amazon, Atlassian, LinkedIn, monday.com, ClickUp, and SigNoz completed the ten largest positive-description counts at those ranks.

This is a volume leaderboard, not a brand-quality ranking. Use it to find repeated lower-rank praise worth reviewing, then compare each brand's wording, prompt mix, and rank rather than its raw count alone.

Named brands praised at fourth or lower

Positive (%) · 10 results

  • monday.com85.97%
  • SigNoz85.59%
  • LinkedIn83.32%
  • Grafana82.88%
  • Alphabet80.09%
  • Microsoft78.85%
  • ClickUp78.75%
  • Amazon78.56%
  • Notion76.6%
  • Atlassian75.56%
  • 0%50%100%
Explore the data (10 rows)
Alphabet6,9348,65880.08781,733
Microsoft5,4396,89878.84891,590
Grafana4,0514,88882.8764196
Notion3,5824,67676.6039462
Amazon3,4664,41278.5585976
Atlassian3,4004,50075.5556382
LinkedIn2,7273,27383.3181358
monday.com2,6603,09485.9729221
ClickUp2,6353,34678.7507291
SigNoz2,3172,70785.5929105

Takeaway

Compare repeated praise with recommendation rank before prioritizing a brand-language change.

The 81.2% result held under three narrower checks

The main fourth-or-lower rate was 81.2%. A June 1 start returned 81.2%, and exact brand identities without parent consolidation returned 81.4%. Counting each ranked brand once instead of each description found positive language on 337,469 of 401,496 fourth-or-lower recommendations, or 84.1%.

The study excluded descriptions that could not be linked to a ranked brand or stable brand identity. It describes observed answer language, not recommendation accuracy, brand quality, buyer opinion, why an engine chose a rank, or causation.

main result
81.2%

640,708 of 788,976

June 1 through July 16
81.2%

640,597 of 788,835

exact brand identities
81.4%

668,015 of 820,417

one count per recommendation
84.1%

337,469 of 401,496

What marketers should do

Positive language appeared in 81.2% of descriptions for brands ranked fourth or lower. The rate was higher at first place, but the size of that gap differed by engine, description form, and industry.

Report recommendation rank and sentiment as separate measures. Audit the same prompts on both engines. Compare like ranks within the same industry. Review specific attributes and positioning claims instead of optimizing for generic praise. Repeat the fixed-window method next quarter before calling a difference movement.

  • Being cited is not enough: how to monitor sentiment in AI answers

How we measured

We analyzed 1,800,746 brand-description assignments across 71,801 brands, 218,189 AI answers, and 16,103 organic prompts on ChatGPT Search and Google AI Mode from May 24 through July 16, 2026.

of descriptions for first-ranked brands were positive
94.1%

367,591 of 390,696

separated first place from fourth or lower
12.9 points

94.1% versus 81.2%

brand-description assignments were analyzed
1.8M

1,800,746 across 71,801 brands

Get the data

Dataset CSVThe metrics behind every figure in this report.

Sources

  1. Semrush: AI Visibility Brand Performance Reports (opens in a new tab) · accessed August 12, 2026
  2. BrightEdge: Google AI Overviews are 44% more likely to criticize brands than ChatGPT (opens in a new tab) · accessed August 12, 2026
  3. G2: The Answer Economy, 2026 AI Search Insight Report (opens in a new tab) · accessed August 12, 2026
  4. Evaluating position bias in large language model recommendations (opens in a new tab) · accessed August 12, 2026

More like this

Does AI praise and criticize the same brand?
AI sometimes praised and criticized the same brand within one answer.
What words does AI use to describe brands?
Most brand descriptions were positive, and much of the praise was shared across brands.
Does AI criticize its top recommendation?
AI occasionally paired its first recommendation with a specific negative claim about the same brand.
How often does a mentioned brand rank first?
Only about one in eight brands mentioned in an AI answer was its actual pick.
Which brands are most associated with each adjective?
Some descriptions had a clear leading brand; others were close contests, including HubSpot and Pipedrive for user-friendly CRM.
What reasons does AI give for its first recommendation?
AI cited reputation more often for first choices than lower-ranked brands. Price was the first choice's most common reservation.

See where you rank and what AI says.

Parse

Index

  • Brands
  • Markets
  • Topics
  • How we measure

For agents

  • Docs
  • Connect
  • Workflows
  • llms.txt

Parse

  • Pricing
  • Research
  • Blog
  • Compare

Company

  • About
  • Contact
  • Corrections

© 2026 Parse, Inc. All rights reserved.

Privacy PolicyTerms of Service