The AI SEO & AEO primer.

Getting your content cited when a user asks a question to ChatGPT, Perplexity, Claude, or Google's AI overviews — instead of just clicking a traditional search result.

AEO overlaps with SEO but optimizes for different signals: clear entity markup, citable statistics, structured FAQs, and passage-level answers. The ranking surface is an LLM, not a SERP.

A reference to your brand, product, or content inside an AI-generated answer — sometimes with a link, sometimes without.

Citation rates are measurable. Run the same prompts across models weekly, track which competitors show up, and engineer the content gaps until you do too.

Publishing thousands of high-quality template pages, each targeted at a specific long-tail query — built from a structured data source, not written one by one.

Executed well (Zapier, Webflow, Clay) it compounds for years. Executed poorly it becomes Google-penalized spam. The difference is source-data quality and per-page usefulness.

Mapping your content to the concepts your ICP cares about, not the individual search phrases. Search engines and LLMs both rank topics, not strings.

An entity is a unit of meaning — "vector database," "MCP server," "PLG motion." We start by enumerating your category's entities and the relationships between them.

A software loop where an LLM reads some inputs, decides what to do, calls tools, reads the results, and either loops or stops.

Different from a single-prompt chatbot. An agent has memory, tool access, and termination conditions. Used well, it replaces rote knowledge work.

An open standard that lets AI agents plug into tools and data sources without custom integrations every time.

Anthropic introduced it; the ecosystem adopted it. Matters because it's how your agents will talk to your CRM, your calendar, your codebase — without bespoke connectors.

Giving an LLM access to your private data at query time so it can cite it in the answer — instead of fine-tuning.

For most B2B use cases, RAG is the right tool. You keep your data out of the training set and can update the knowledge base without retraining.

Whether your email actually lands in the inbox instead of spam. The single most under-engineered part of AI outbound.

SPF, DKIM, DMARC, warmup schedules, inbox rotation, sender reputation. Most "AI SDR" tools skip half of this and blame the copy when replies tank.

Triggering outbound when something real happens — funding round, new hire, tech-stack change, product launch — instead of blasting a static list.

Reply rates go up 3–5× when the timing is right. The engineering work is in wiring enough signal sources and writing the copy that references them plausibly.

A specific, numerical description of the accounts most likely to buy, stay, and expand. Not a persona. Not a vibe.

Derived from your actual closed-won data: employee count ranges, industry codes, tech-stack signals, funding stage, hiring patterns. If you can't score a prospect in 30 seconds, it isn't an ICP yet.

Google's three technical measurements of how fast and stable your site feels. Affects ranking and conversion.

Largest Contentful Paint, Interaction to Next Paint, Cumulative Layout Shift. Boring to fix, measurable to improve, and almost every AI SEO client needs the work.

Invisible metadata on your pages that tells search engines and LLMs exactly what each piece of content is — a review, a price, a FAQ, a how-to.

One of the highest-leverage AEO levers. LLMs lean on structured data to answer confidently. Done right, every page on your site is machine-readable.