Creates a clean starting line. You need a baseline crawl and evidence of current AI exposure before touching anything. Without it you can’t prove later that the work moved the needle.
| Step | Detail & Rationale |
|---|---|
| Project board setup | A Kanban (Asana, Jira, Trello) ensures the 40‑plus subtasks stay visible. Add columns for To‑Do / In‑Progress / QA / Done. |
| Baseline crawl | Run Screaming Frog + server log sampling. Export: • Indexable URL list (gives task scope) • Status‑code distribution (find errors) • Existing structured‑data types (so you know coverage gaps) |
| AI visibility screenshots | Search Google SGE, Perplexity, ChatGPT‑4o w/ browsing, Claude. Capture results for: • Brand query (“Who is Brand X?”) • Category (“best waterproof jackets under $200”) • Top 5 product names. This “before” evidence is vital for quarterly reporting and debugging. |
Almost every AI platform still relies on a crawlable, fast, mobile‑friendly site. If Googlebot or GPTBot can’t load a page, your brilliant content never makes it into training data or live retrieval.
robots.txt hygiene
Goal: maximise access while retaining security.
Action: explicitly Allow Googlebot, Bingbot, GPTBot, PerplexityBot. Keep a daily‑rotating log of 404/403 hits from those user‑agents to catch accidental blocks.
XML sitemaps
Goal: shorten discovery time for new URLs.
Action: generate separate sitemaps for /products, /blog, /images, list in sitemap_index.xml, ping GSC after every deployment.
Status‑code clean‑up
Goal: remove crawl waste and canonical confusion.
Action: 301 chains → single hop; flag non‑canonical duplicates for merge; ensure all variant URLs point to one canonical (e.g., /blue-shirt?size=m → /blue-shirt/).
Core Web Vitals
Goal: pass LCP < 2.5 s, CLS < 0.1.
Why AI cares: Google’s real‑time index quality scoring downgrades “slow” URLs for inclusion in SGE answers. Perplexity’s crawler also deprioritises very heavy pages to save compute.
Mobile‑first parity
Action: Render page in Chrome DevTools iPhone viewport; ensure no hidden specs or schema only visible on desktop. AI crawlers use smartphone user‑agent by default.
Generative engines look for machine‑readable signals first, free‑text second. Correct schema acts like a data API that AI can trust.