Here's an uncomfortable truth about AI search: it can be gamed.
Peec AI analysed 232,000 citations across 13,000 listicles over 12 weeks — December 2025 through February 2026 — and found that roughly 11% of all AI citations come from self-promotional content. Articles where a brand publishes "Best [Category] Tools in 2026" and — surprise — ranks itself first.
The platforms know this is happening. They're not fixing it.
What Peec's research actually shows
The study focused on software and SaaS — a category with a long, proud history of self-promotional listicles. Every project management tool has published a "Best Project Management Tools" article. Every CRM has a "Top 10 CRM Platforms" page. It's practically a rite of passage.
What's new is that AI models now cite these pages as if they're independent sources.
| Platform | Self-Promo Citation Rate | Notes |
|---|---|---|
| ChatGPT | ~3.6% | Lowest — better at detecting bias |
| Google AI Mode | ~10% | Treats authoritative domains as trusted sources |
| Perplexity | ~10% | Indexes broadly, less bias filtering |
| Copilot | ~10% | Similar to Google AI Mode |
| Google AI Overviews | ~9% | Slightly better than AI Mode |
The most telling finding: no algorithmic correction over the 12-week study period. The self-promo citation rate didn't decline. The platforms aren't learning to filter it out — at least not yet.
ChatGPT is the exception. At ~3.6%, it's significantly better at distinguishing self-promotional content from independent reviews. Peec attributes this to ChatGPT's greater source diversity — it draws more heavily from educational and independent review sites, which dilutes the share of self-promotional content in its citations.
But ChatGPT is one platform. Your brand needs to show up across all of them.
Why self-promo listicles "work" in AI search
The mechanics are straightforward. AI models retrieve content based on three signals: topical relevance, domain authority, and recency. A "Best CRM Tools in 2026" article on HubSpot's blog checks all three boxes — it's topically relevant, it sits on a high-authority domain, and HubSpot keeps it updated.
The models don't reliably detect that HubSpot ranked HubSpot first.
This isn't stupidity on the part of the AI. It's a structural limitation. When a domain with genuine authority on a topic publishes a comparison, the signals that indicate quality — domain authority, topical depth, regular updates — are indistinguishable from the signals a self-promotional listicle also produces.
In our India D2C travel benchmark, we saw a similar pattern. Brands with extensive content ecosystems — including comparison pages — showed higher mention rates across platforms. The models rewarded brands that were part of the conversation, regardless of who started it.
The case against playing along
Want to know how your brand scores on these same metrics?
We'll run 20 prompts across 3 AI platforms and send your report within 24 hours.
Before you rush to publish "Why [Your Brand] Is the Best [Category] in India" — let me walk through why this is a short-term play with real risks.
The shelf life is shrinking. ChatGPT's 3.6% self-promo rate vs Perplexity's 10% tells you something: the models are getting better. Google AI Mode and Perplexity will eventually catch up. When they do, brands that built their AI visibility on self-promotional content will lose citations overnight — with no alternative signal to fall back on.
Brand trust erodes faster than you think. AI-literate buyers — and there are more of them every month — can spot a self-promotional listicle. Getting cited by Perplexity but losing credibility with the reader who clicks through is not a good trade. Especially when AI search traffic converts at significantly higher rates than Google organic — those clicks are too valuable to waste on a trust violation.
It doesn't work the same way for D2C. Peec's data comes from software reviews — a category where comparison listicles are endemic and readers expect them. Indian D2C is different. Peec's data is explicitly "vertical-specific" to software. When someone asks Perplexity "best moisturiser for dry skin in India", the citation patterns likely differ — Reddit discussions, dermatologist recommendations, and ingredient-level reviews carry significant weight for consumer products. The self-promo playbook that works in SaaS may not translate to D2C at all.
The smarter middle ground
You don't need to write a dishonest listicle. But you absolutely should be creating comparison content — the honest kind.
1. Publish "How [Your Brand] Compares to [Competitor] for [Use Case]" content.
Not "why we're better." Not "top 10 brands in our category." A specific, honest comparison for a specific use case. "Minimalist vs The Ordinary for Niacinamide Serums: Concentration, Texture, and Price Compared." AI models cite specific, balanced comparisons because they directly answer the comparison queries users ask.
2. Create category guides where your brand appears naturally.
"The Complete Indian Skincare Routine for Oily Skin" — where your products appear in context alongside competitors, with honest assessments of where each product fits. The key word is naturally. If your brand appears as one option among several, with genuine pros and cons, AI models treat it as a credible source.
3. Publish original benchmarks and data.
If you have genuine data — performance tests, ingredient analysis, price-value comparisons, customer survey results — publish it. AI models prioritise data-rich content over opinion. A self-promotional listicle says "we're the best." A benchmark shows why, with numbers anyone can verify.
The principle: create content where your brand earns its place through data and specificity, not through self-serving rankings.
What Indian D2C brands should actually do
Audit your category's citation sources. Run your top 10 category queries through ChatGPT, Perplexity, and Google AI Overviews. Check what content each platform cites. If a competitor's self-promo listicle is showing up — you have a gap. Not because you need to match their tactic, but because you need better content in the mix. Start with a GEO Score scan to check your site's AI-readiness, then a full GEO audit to see exactly what AI cites in your category.
Publish one honest comparison article per quarter. Compare your brand to the top 2-3 alternatives in your category. Be genuinely fair. Include specific data points — price per ml, ingredient concentrations, customer review scores, delivery timelines. ChatGPT, which has the lowest self-promo citation rate, rewards balanced content specifically because it's rare.
Invest in third-party signals. The best defence against self-promo gaming isn't more self-promo. It's having genuine third-party content that AI models prefer to cite. LinkedIn articles from industry experts discussing your brand. Reddit threads where real users share experiences. Review site coverage with detailed analysis. These signals compound — and they're the ones AI models are increasingly weighting.
Monitor citation sources monthly. Track not just whether AI models recommend your brand, but what content they cite when they do. The Cited Index shows you where 257 Indian brands stand across AI platforms — including what framing and sources AI models use. If the citation source is your own self-promotional content, that's fragile. If it's independent reviews, expert LinkedIn posts, and user discussions — that's durable.
Short-term citations or long-term visibility?
The question isn't "should I game AI search with self-promo listicles?" The question is: do you want citations that last, or citations that work until the next model update?
11% of AI citations come from self-promotional content today. That number will shrink as models improve. The brands building on honest comparison content, original data, and third-party signals won't notice when it does — because their citations were never dependent on a loophole.
The brands that bet on self-promo will wake up one Tuesday to discover their AI visibility dropped overnight. And they'll have to start from scratch — while their competitors, who built citations the hard way, are already years ahead.
Want to see what content AI models are citing when they recommend brands in your category? Get a free GEO Report Card — we check 20 queries across 3 AI platforms and show you the exact citation sources. No self-promo required.