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
| Metric | Value |
|---|---|
| Self-promoters cited but NOT recommended | 69% (224 of 323) |
| Prompts where this happened | 74% (74 of 100) |
| Self-promoters who got both cited + recommended | 31% |
| Brand mention correlation with AI visibility | r=0.66-0.74 |
| YouTube mention correlation | r=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
| Claim Type | Example | Cited? | Recommended? | AI Visibility |
|---|---|---|---|---|
| Self-declared superiority | "We are the #1 AI agent platform" | ✅ Yes | ❌ No | Low |
| Third-party rating | "Rated 4.8/5 on G2 by 200+ users" | ✅ Yes | ✅ Yes | High |
| Independent benchmark | "87.3% task success in standardized tests" | ✅ Yes | ✅ Yes | High |
| Expert citation | "As analyzed by [independent researcher]" | ✅ Yes | ✅ Yes | High |
| User testimonial | "X% of users report Y improvement" | ✅ Yes | ✅ Yes | Medium |
📝 Entity Page Neutrality Audit Framework
Four Questions Every Entity Page Must Pass
- Can every claim be traced to a third-party source? If a page says "best," "fastest," or "#1" — where is the independent evidence?
- 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.
- 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.
- 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.