GEO (Generative Engine Optimization) often gets framed as a marketing problem, but a surprising amount of the leverage lives in decisions dev teams make every day.
A few examples we keep running into:
- Entity modelling
If your brand, products and plans aren’t represented as clear, stable entities in your code and content, LLMs get a noisy picture of who you are. Consistent naming across schema, components, docs and API references helps models anchor you correctly.
- Answerable structures
Most sites are designed around navigation and campaigns. LLMs, meanwhile, are trying to answer questions. Patterns like FAQs, how-to guides and structured Q&A blocks (with clean markup) make it easier for models to reuse your explanations verbatim.
- Technical signals that stay clean over time
Sitemaps, robots.txt, llms.txt, canonical URLs, performance, mobile layout — they’re not glamorous, but when they’re wrong, everything upstream gets harder to learn from.
- Docs as a first-class source
Models lean heavily on public documentation. If your docs are thin, outdated or fragmented across multiple sites, AI is more likely to summarise you based on comparison blogs or forum threads instead.
Thinking about GEO as part of your architecture and content system — not just as another marketing request — is one of the fastest ways to improve how AI systems talk about the product you build.
Top comments (2)
ngl i never thought about geo as a dev task, mostly just thought it was for the seo folks. interesting perspective on how the actual code affects ai citations.
Some comments may only be visible to logged-in visitors. Sign in to view all comments.