Harvard Business Review just published "LLMs Are Overtaking Search" by Graham Kenny and Ganna Pogrebna.
The framing is right. The urgency is real. The tactical guidance is thin.
HBR is a strategy publication — that's expected. But if you're a marketing manager or SEO lead who finished that article and thought "OK, how do I actually do this?" — this post is for you.
We run GEO audits. We've analyzed how AI systems actually decide what to cite. Here's the implementation layer that HBR didn't have room for.
What HBR Got Right
The article identifies three strategic shifts:
Branding → AI recall
Being findable in search isn't the same as being cited in AI answers. AI systems favor brands they've seen repeatedly across credible sources.
PageRank → Citation authority
The signals that drive Google rankings (backlinks, domain authority) partially overlap with AI citation signals — but the correlation is weaker than most people assume.
Paid → Earned presence
AI answers don't include ads. Your paid media budget contributes zero to your AI visibility. Earned editorial coverage does.
These are accurate. But "build citation authority" and "increase earned presence" are directional, not actionable. Here's what those actually mean in technical practice.
The Tactical Translation
Shift 1: "Build AI Recall" = Fix Your Technical Access First
Before any content strategy matters, AI crawlers have to be able to read your site.
A surprising number of sites accidentally block GPTBot (ChatGPT), ClaudeBot (Anthropic), PerplexityBot, and OAI-SearchBot through wildcard Disallow: / rules in their robots.txt files. If those bots can't crawl your site, you cannot appear in AI-generated answers — regardless of how good your content is.
Check your robots.txt right now: yourdomain.com/robots.txt
Look for a wildcard block:
User-agent: * Disallow: /
This blocks everything. Add explicit Allow rules for AI crawlers:
User-agent: GPTBot Allow: / User-agent: ClaudeBot Allow: / User-agent: PerplexityBot Allow: / User-agent: OAI-SearchBot Allow: /
This is a one-hour fix that eliminates a hard blocker for every piece of content on your site.
Second step: publish an llms.txt file at yourdomain.com/llms.txt. This is the emerging standard (similar to robots.txt) for telling AI systems how to interpret your site. Include your company description, key service/product categories, and links to your most authoritative pages.
Shift 2: "Build Citation Authority" = Three Specific Content Signals
HBR correctly identifies that citation authority differs from link authority. Here's what the academic research says about what actually drives it.
The Princeton/Georgia Tech GEO paper (KDD 2024) measured exactly which content changes boost AI citation rates:
| Content change | Measured lift in AI citations |
|---|---|
| Answer-first opening paragraph | +67% citations |
| Statistics with named sources ("According to Gartner...") | +30–40% citations |
| Expert quotations with credentials | +37% citations |
| FAQ schema markup | 3.2× citation rate in Google AI Overviews |
| Data tables vs. equivalent prose | 4.1× more AI citations |
These aren't estimates — they're measured outcomes from the Princeton/Georgia Tech study.
The HBR article mentions "original data and named proprietary frameworks." That's right, but it's one tactic in a larger system. The more actionable principle: make your content machine-extractable.
AI systems extract specific answers from your content in milliseconds. Content that loads slowly due to JavaScript rendering, buries its key points in prose, or lacks structured metadata is harder to extract — and less likely to be cited, even if it's high-quality.
Practical implementation:
- Rewrite article intros to answer the query in the first sentence, not the third paragraph
- Add FAQPage schema to every key page (one JSON-LD block, 2–3 hours of work, measurable lift within 3–6 weeks)
- Name your data sources in every claim ("BrightEdge Q3 2025 data" not "recent research")
- Put key comparisons in tables, not bullet lists — tables are cited 4.1× more often
Shift 3: "Earn Presence" = Specific Platforms and Mechanisms
HBR is right that editorial coverage beats on-site optimization for AI visibility. Here's the channel-specific breakdown:
Reddit: AI systems heavily weight Reddit for community validation of claims. A thread where practitioners discuss your approach is more citable than your own blog post making the same argument. The mechanism: Reddit content is structured (thread → comments), attributed (usernames = perceived authenticity), and machine-readable. A subreddit post that earns substantive replies becomes an AI citation source.
YouTube: YouTube transcripts are now machine-readable and weighted heavily by AI systems. YouTube now appears in 16% of LLM answers vs. 10% for Reddit (Adweek, January 2026). If your methodology can be explained on video, the transcript is a new GEO asset.
Named experts cited by others: The HBR article mentions "credentialed expert association." The mechanism in practice: when an expert with verifiable credentials (published work, institutional affiliation, professional titles) cites your data or quotes your team, that citation is more likely to be extracted by AI systems. Anonymous "staff writer" content rarely appears in AI citations.
Platforms that don't help much: Paid social, email, most gated content, PDFs without accessible text layers. These don't contribute to AI citation authority regardless of their SEO or paid media value.
The Three Things to Do This Week
If you read the HBR article and want to act:
Three high-impact actions
- Audit your robots.txt — 30 minutes. Check if you're accidentally blocking AI crawlers. If yes, fix it immediately. This is a hard blocker — no content strategy matters until crawlers can read your site.
- Add FAQPage schema to your top 5 pages — 2–4 hours. This is the highest-leverage single technical change for Google AI Overviews. The CXL empirical study found a 3.2× citation lift from FAQPage schema alone.
- Run one answer-first content rewrite — Pick your most important page. Rewrite the intro so the first paragraph directly answers the primary search query. Princeton/Georgia Tech measured a 67% citation increase from this change. Measure citation rates in 4–6 weeks.
What a Full GEO Audit Covers
If you want to know where you stand across all five AI platforms (ChatGPT, Perplexity, Google AI Overviews, Claude, Gemini), a GEO audit covers:
- Technical access — robots.txt, JS rendering blindness, llms.txt
- Structured signals — schema markup completeness and accuracy
- Content architecture — extractability, answer-first structure, source citation
- Brand mention presence — across all major AI platforms
- Gap analysis — vs. the top 3 competitors being cited in your category
The HBR article is right about the direction. The audit tells you exactly where you are and what to do first.
Get your free GEO audit
We'll analyze your site against every AI citation signal and tell you exactly what to fix first.
- Technical access audit (robots.txt, crawler access, JS rendering)
- Schema markup assessment (FAQ, Organization, Article)
- Brand mention baseline across ChatGPT, Perplexity, Gemini, Claude
- Prioritized fix roadmap — highest citation impact first
Frequently Asked Questions
What did the HBR article say about LLMs overtaking search?
The Harvard Business Review article "LLMs Are Overtaking Search" (Kenny & Pogrebna, March 2026) identifies three strategic shifts: (1) branding is now about AI recall, not just search findability; (2) PageRank-style link authority is weakening as an AI citation signal; (3) paid media contributes nothing to AI visibility while earned editorial coverage does. The article correctly frames the directional shift but does not provide implementation-level tactics for how to actually improve AI citation rates.
How do I get my brand cited in AI search results like ChatGPT and Perplexity?
To get cited in AI search results: (1) Check your robots.txt is not blocking GPTBot, ClaudeBot, PerplexityBot, or OAI-SearchBot. (2) Publish an llms.txt file at your domain root. (3) Rewrite article intros so the first paragraph directly answers the primary query — answer-first structure increases AI citations by 67% per Princeton/Georgia Tech research. (4) Add FAQPage JSON-LD schema — CXL's empirical analysis found a 3.2× citation lift. (5) Name your data sources explicitly in every claim. (6) Put key comparisons in tables — tables are cited 4.1× more than equivalent prose.
What is the difference between SEO and GEO?
Traditional SEO optimizes for search engine rankings using signals like domain authority, backlinks, and keyword relevance. GEO (Generative Engine Optimization) optimizes for AI citation — making your content extractable and citable by systems like ChatGPT, Perplexity, Google AI Overviews, and Claude. The key difference: 88% of AI citations bypass the conventional organic top-10, meaning high Google rankings do not reliably translate to AI visibility.