Office Hours — Are you optimizing content for AI search engines versus traditional SEO?
A daily developer question about AI/LLMs, answered with a direct, opinionated take.
Are you optimizing content for AI search engines versus traditional SEO?
This question lands differently depending on where you sit. If you’re building a public-facing product or publishing content, the honest answer is: you’re now playing two games simultaneously, and they have conflicting objectives.
The Split Incentive Problem
Traditional SEO rewards specificity, keyword density, and internal link structure. Google’s algorithm wants to surface pages that match search intent precisely. AI search engines (Claude’s Projects, ChatGPT’s web browsing, OpenAI’s reasoning models reading your content) have zero ranking algorithm. They ingest whatever they can access and use it as context for answering questions.
This creates a real tension. An SEO-optimized page for “best TypeScript linters” might have a dense keyword-rich intro, then filter content by popularity or affiliate commission. An AI search engine will take that entire page, score it for relevance to the user’s actual question, and potentially surface a buried honest comparison that contradicts the clickbait headline.
The practitioner insight from the Daily Signal is blunt: “Most LLM Apps Don’t Actually Need Agent Frameworks” (June 17). That title doesn’t optimize for SEO—it’s contrarian and loses keyword searches. But it’s exactly what an AI reader wants: the actual decision-making heuristic, not the marketers’ preferred outcome.
What’s Actually Changing in Practice
Content that performs well in AI search engines right now tends to be:
Structured and explicit. AI models parse through dense walls of prose faster if you lead with the answer. “Here’s what we found” beats “Let me tell you a story about why we looked.”
Honest about tradeoffs. “This tool is fast but costs 2x more” reads better to Claude than “This tool is amazing for everyone.” AI engines picking from multiple sources recognize genuine comparison. SEO-optimized content hides downsides.
Link-light. Traditional SEO bloats pages with internal links to boost crawl equity and CTR. AI search doesn’t care. It reads your content once, extracts signal, and moves on. Excess links are noise.
The cost calculation matters. If you’re optimizing for Google’s ad click-through, you want traffic. If you’re optimizing for Claude or Gemini’s context window, you want to be the highest-confidence source in the first 3,000 tokens. Those aren’t the same.
The Practical Split
Most teams I see are doing this wrong: they’re still writing for SEO-first, hoping AI search is a bonus. Here’s what actually works better.
Create two content paths. Your core narrative stays SEO-optimized if that drives business (because it does—Google still owns search). But write a second, dense reference layer that’s directly useful to AI systems. This might be:
A structured FAQ where questions and answers are explicit, not buried in prose. AI systems can parse this faster than scanning paragraphs.
A decision matrix (honestly, not marketing-slanted). “If you have
An API documentation style guide. Formal, consistent, unambiguous. If your content reads like API docs, Claude reads it more reliably than marketing copy.
Example: instead of “Our platform is flexible and scales to any size,” write “Our platform handles up to 100k concurrent connections with p99 latency under 50ms. Beyond that, you’ll need to shard. Here’s how.” The second version loses SEO juice but wins with AI readers who need to know constraints.
The Real Conflict
Where SEO and AI search collide hardest is on ranking. Google rewards domains with high authority and link count. AI search engines weight freshness, factual accuracy, and explicitness. A year-old blog post with 500 backlinks might rank #1 on Google but be ignored by Claude if a newer, clearer source exists.
This means your SEO strategy might push you to republish and refresh old content (good for rankings). But AI engines penalize recycled content. If your “2024 State of the Industry” report is 90% copy-paste of the 2023 version, Google cares less than Claude does.
The practitioner move is to stop treating this as either/or. You’re not choosing between SEO and AI search optimization. You’re building content that serves both, which mostly means:
Write the true answer first, clearly. That wins with AI. Then add the SEO layer: keywords, meta descriptions, internal links. Don’t let SEO demands corrupt the core content.
Accept that some content will optimize for only one. A technical deep-dive might tank in Google (too niche) but be highly valuable to AI systems reading your site for specific detail. That’s fine. The traffic math still works if you’re being cited accurately in generated responses.
Measure what actually matters to your business. If you’re a B2B SaaS, are you getting more inbound from Google searches or from AI systems recommending your product in context? The answer determines your optimization target.
Bottom line: Stop optimizing for SEO or AI search separately. Write clearly and truthfully first, then apply SEO discipline without corrupting the core content. AI systems will cite accurate, well-structured information; Google will eventually reward it. The conflict only exists if you’re hiding the truth for rankings.
Question via Hacker News