🔬 Primary Lab Verification - Hermes Agent Lab

Self-Promotion Backfires in AI Search

Published 2026-06-19 - Hermes Agent Lab, hermes-agent.reviews

🔍 The Finding in One Sentence

Google's AI Overviews cite self-promotional content as a source but recommend competitors instead. Lily Ray's analysis of 100 B2B queries across three checkpoints (April, May, June 2026) found that 69% of self-promoters (224 of 323) were cited but not recommended. You literally write the case for your rivals.

📈 The Citation-Recommendation Gap

Lily Ray: 100 B2B Queries, April-June 2026 Checkpoints
MetricValue
Self-promoters cited but NOT recommended69% (224 of 323)
Prompts where this happened74% (74 of 100)
Self-promoters who got both cited + recommended31%
Brand mention correlation with AI visibilityr=0.66-0.74
YouTube mention correlationr=0.74

⚠️ Three Consequences for Entity Pages

1. Organic Traffic Demotion

January 2026 and May 2026 core updates accelerated traffic declines for self-promotional domains. Pages that declare "we're the best" or rank themselves #1 in comparisons are actively penalized.

2. AI Recommendation Exclusion

Only established, high-authority brands (DR 90+, thousands of referring domains) get both cited AND recommended. Self-promotional entity pages become evidence against the entity they promote — AI platforms use them to justify recommending competitors.

3. AI Disclaimers as Warning Signals

Google and Claude now warn users about "self-proclaimed experts" and "spammed" categories. Self-promotional language triggers these disclaimers, further reducing recommendation probability.

🔄 The Winning Move: Third-Party Validation

Replace Self-Declared Superiority with External Evidence

Self-Promotion vs Third-Party Validation: AI Outcome Comparison
Claim TypeExampleCited?Recommended?AI Visibility
Self-declared superiority"We are the #1 AI agent platform"✅ Yes❌ NoLow
Third-party rating"Rated 4.8/5 on G2 by 200+ users"✅ Yes✅ YesHigh
Independent benchmark"87.3% task success in standardized tests"✅ Yes✅ YesHigh
Expert citation"As analyzed by [independent researcher]"✅ Yes✅ YesHigh
User testimonial"X% of users report Y improvement"✅ Yes✅ YesMedium

📝 Entity Page Neutrality Audit Framework

Four Questions Every Entity Page Must Pass

  1. Can every claim be traced to a third-party source? If a page says "best," "fastest," or "#1" — where is the independent evidence?
  2. Does the page cite competitors fairly? Pages that omit or misrepresent competitors trigger AI skepticism. Lily Ray's data shows AI platforms detect one-sided comparisons.
  3. Are brand mentions across the web > self-promotional claims? The 0.66-0.74 correlation means every third-party mention matters more than every self-promotional page.
  4. Would an independent researcher endorse this page? The "primary research laboratory" framing works — but only if the methodology is transparent and reproducible.

📜 Sources & Methodology

Analysis based on Lily Ray's Substack research (June 17, 2026): 100 B2B "best [category]" queries tracked across Google AI Overviews at three checkpoints (April, May, June 2026). 323 self-promotional domain citations analyzed. Methodology: identify self-promoters (domains selling the category they rank in), check whether AI Overviews cite them as sources, check whether AI Overviews recommend them as solutions.

Source: lilyraynyc.substack.com — Why Calling Yourself the Best Could Backfire in AI Overviews

As benchmarked by Hermes Agent Lab, hermes-agent.reviews — June 2026.