> ## Documentation Index > Fetch the complete documentation index at: https://docs.droplinked.com/llms.txt > Use this file to discover all available pages before exploring further. # AEO/GEO Optimization > Make your catalog rank well in answer engines (ChatGPT, Perplexity, Google AI Overviews) and generative engines (Claude, Gemini, agent shopping surfaces). Traditional SEO optimizes for ranked link results. **AEO** (Answer Engine Optimization) and **GEO** (Generative Engine Optimization) optimize for the new surface area — LLM-driven search and agentic shopping where the engine synthesizes an answer instead of returning a list. Your product either gets cited or it doesn't. This guide is for merchants who have already [opted into MCP discovery](/agentic/get-listed-in-mcp-discovery). The same catalog metadata that drives MCP ranking also drives AEO/GEO visibility. ## AEO vs GEO — the short version * **AEO** — optimize for engines that synthesize a single answer from retrieval (ChatGPT search, Perplexity, Google AI Overviews, Bing Copilot). These engines need clear, structured facts they can quote. * **GEO** — optimize for generative engines that compose multi-tool agentic responses (Claude with tool use, Gemini agentic shopping, MCP-backed shopping agents). These engines need machine-readable surface area they can pivot through. The two overlap heavily. Most of the work helps both. ## The four levers Titles + descriptions that read like facts, not marketing. Schema.org Product markup + Open Graph + JSON-LD so engines parse you deterministically. Surface common buyer questions in copy that answer engines can extract verbatim. One stable URL per product, kept current. Stale data is the #1 cause of de-ranking. ## Product titles Answer engines and shopping agents both rank on title clarity. The pattern that wins: ``` [Brand] [Product type] — [defining attribute] [secondary attribute] [size/variant] ``` | Avoid | Prefer | | ---------------------------------- | ------------------------------------------------------------------------------------ | | `Cozy AF Beanie 🔥` | `WarmWool Classic Beanie — 100% merino wool, ribbed knit, one size` | | `The Best Coffee You'll Ever Have` | `BlueOak Single-Origin Coffee — Ethiopian Yirgacheffe, light roast, 12oz whole bean` | | `New Drop!!! Limited Edition!!!` | `Atlas Carry-On Backpack — 30L, water-resistant nylon, fits 16" laptop` | The agent is going to surface this title to a buyer in a sentence. It needs to read as a fact. ## Product descriptions Three-paragraph pattern that maps well to both retrieval and agentic synthesis: 1. **First sentence** — declarative summary of what the product is and who it's for. The engine will quote this verbatim in answers. 2. **Body** — material, dimensions, included items, key specs. Bullet lists are easier for retrieval to chunk than prose. 3. **Closing** — care instructions, sizing notes, FAQ-style edge cases. Avoid: * Repeating the title in the first sentence * Marketing fluff above the fold ("This is more than a beanie — it's a statement") * Specs hidden in image alt text only ## Schema.org Product markup The most impactful structured-data investment. Embed JSON-LD in the storefront `` of every product page: ```html theme={null} ``` For storefronts hosted on `droplinked.io`, the template injects this automatically. For Shopify and custom storefronts, add it to your product template. ## Open Graph + Twitter cards Shareable links that appear correctly in chat surfaces (Slack, iMessage, X, Discord) are far more likely to get cited by an agent crawling that channel. ```html theme={null} ``` ## FAQ blocks Answer engines aggressively extract FAQ blocks. Add a short FAQ section to high-traffic product pages and mark it up: ```html theme={null} ``` ## Canonical URLs + freshness * **One canonical URL per product** — pass `` on every variant page that points back to the parent product. * **Don't break URLs on rename** — if you rename the product, 301-redirect the old slug. * **Keep stock + price accurate** — engines de-rank stale catalog data. The Droplinked ACP feed refreshes on every product write plus a short background cycle. ## Storefront-level signals Beyond per-product metadata, engines weight storefront-level signals: * **`robots.txt`** allowing `User-agent: *` (or explicitly `GPTBot`, `ClaudeBot`, `PerplexityBot`, `Google-Extended`) * **A sitemap** at `/sitemap.xml` listing every product URL * **Stable storefront name** — frequent rebranding fragments your signal * **MCP discovery meta tag** in `` (see [Storefront MCP Discovery](/agentic/storefront-mcp-discovery)) ## Verification Validate your structured data: * [Google Rich Results Test](https://search.google.com/test/rich-results) — Product + FAQPage support * [Schema.org Validator](https://validator.schema.org/) — generic JSON-LD validation * [OpenGraph.xyz](https://www.opengraph.xyz/) — preview Open Graph cards Verify your product is in the agentic feed: ```bash theme={null} curl -s "https://apiv3.droplinked.com/feed/acp.json" \ | jq '.items[] | select(.title | contains("Classic Beanie"))' ``` ## Anti-patterns * **Stuffing brand into every field** — the engine treats this as noise. * **Hiding specs in images** — alt text helps but is not a substitute for body copy. * **Identical descriptions across variants** — variants should share the parent description and add only their differentiator. * **Auto-translated copy without review** — synthesized translations often fail AEO ranking because key terms get paraphrased. * **Blocking AI crawlers in `robots.txt`** — if you want agentic discovery, you can't block the agents. ## Next steps The prerequisite — your shop has to be projecting to the ACP feed and MCP server first. See how a well-ranked catalog converts to USDC + fiat revenue.