Common Ai Readiness mistakes that still ship

I still reach for curl first. Robots file, sitemap headers, a peek at the homepage title. Fine for a first look. The mistake is stopping there and calling the site ready for answer engines.

SlaySlop's glossary calls the check Answer-engine readiness. It "reviews signals that help AI crawlers access and understand public content." The failures below are the ones that still ship. None of them require a research paper. They require someone to stop treating a chatbot demo as a crawl audit.

Outcome if you avoid these

AI crawlers can reach the routes that matter and find enough stable context to understand them. Your report stays about public signals, not about whether a single assistant felt friendly on Tuesday.

Boundaries

Read-only scans. Permission required. SlaySlop does not edit the site. Answer-engine readiness is one SEO and AEO check among others: titles, meta descriptions, structured data, robots, sitemaps, social tags. Closing a readiness finding does not mean the rest of SEO is done. It also does not mean any specific model will cite you.

Mistake 1: Staging robots rules in production

Someone leaves Disallow: / or a broad /docs block from preview. Humans with the bookmark still browse. Crawlers politely leave. I have watched a launch week burn a day on "why isn't AI picking this up" when the file was doing exactly what it said.

curl -sL "https://www.example.com/robots.txt" | sed -n '1,80p'

If the paths you care about are disallowed, fix the file before you commission more content. SlaySlop's SEO flow checks robots rules in the same pass as the rest of the search signals. Believe that finding.

Mistake 2: Sitemap theater

A <link rel="sitemap"> or robots Sitemap: line points at a URL that 404s, returns HTML, or lists last month's preview host. Or the sitemap is perfect and the important URL is neither linked nor listed. Discovery fails quietly.

# Bad: advertises a map that does not resolve
Sitemap: https://www.example.com/sitemap.xml
# curl -sI that URL → 404

The catalog's Sitemap check is about discovery and reachability. Reachability is not optional.

Mistake 3: Titles that only make sense beside a logo

Home, Welcome, ClientName on every primary route. Sighted users get the brand from the header image. An answer engine gets a string that could belong to anyone. Meta descriptions missing or duplicated across twenty pages make the understanding side worse.

<!-- Ships too often -->
<title>Home</title>

<!-- Boring, usable -->
<title>Read-only website scans — Example</title>
<meta name="description" content="Public scans for security, SEO, and performance with ranked findings and evidence." />

SlaySlop's product copy says indexability and missing context matter more than low-impact metadata tweaks. Fixing a useless title is missing context. Spinning synonyms is the low-impact tweak.

Mistake 4: Trusting the template, not the render

The React tree has a helmet title in source. The shipped shell sends empty or default head tags until hydration, and the crawler that matters never sees the hydrated version the way you do in a logged-in session. Or the canonical points at localhost from an env default.

I open the rendered DOM. Then I fetch headers. Then I argue. SlaySlop's SEO product notes that browser execution catches signals that only appear after JavaScript runs. That line exists because this mistake is common.

Mistake 5: Structured data that lies

JSON-LD that describes a Product with a price the page does not show, or an Organization block copied from another brand. That is not "AI readiness." That is contradictory public context. Match visible content. If you cannot keep it honest, omit it until you can.

Mistake 6: Chatbot theater as proof

Someone pastes the URL into an assistant, gets a half-right summary from training data or a cached snippet, and marks AEO done. That demo does not verify robots access, sitemap reachability, or on-page context. It verifies that a model can talk. Different job.

Optional files like llms.txt can help when you maintain them—SlaySlop publishes its own—but they are not a substitute for reachable pages with clear metadata. Do not invent a requirement that every site must have one; the glossary bar is broader: signals that help access and understanding.

Mistake 7: Fixing copy before fixing blockers

Rewriting the blog while pricing is noindexed is backwards. The SEO and AEO product order is deliberate: crawl, inspect signals, check AI readiness, fix blockers first. I still catch myself polishing paragraphs when the route is blocked. Curl the constraints first.

Happy path I actually use

Permissioned scan on production or a public staging host the client will ship. Read Answer-engine readiness findings next to robots, sitemap, title, description, and structured data. Open each affected URL. Change one class of issue at a time. Rescan.

# Minimal evidence pack before blaming "the algorithm"
HOST="https://www.example.com"
curl -sI "$HOST/robots.txt" | head -n 15
curl -sI "$HOST/sitemap.xml" | head -n 15
curl -sL "$HOST/" | rg -n "<title>|description|canonical|ld\\+json" | head

When SlaySlop ranks findings by impact, start at the top. High-impact access and context problems should not wait behind social-preview tidy-ups.

Related next step

Keep Answer-engine readiness on the same triage board as the rest of SEO and AEO. The definition is on Answer-engine readiness. If you want the public surface checked in one pass with evidence beside each issue, run a SlaySlop scan on a URL you are allowed to test.