shop-hanghieu.com

Of everything your server sends for this page, very little is actual readable text. The rest is code, styling, and markup wrapper. Agents fetched a lot of bytes and found few words.

Agents work with retrieval budgets. A page that is 2% text either gets skimmed (and mis-summarised) or skipped. More signal per byte means more of your actual message survives into the agent’s summary.

1. Confirm your real copy is server-rendered. See the JavaScript check; these two usually fail together.
2. Cut boilerplate wrappers: deeply nested divs, inline SVG logos repeated per section, base64 images inlined into HTML.
3. Move large inline styles and scripts into external files.
4. Check how much of the document is inline script: your report now says. If most of it is, that is your framework serializing state for hydration, often a second copy of text already in the HTML. Send only the props a component cannot recompute, and render statically where a page does not need to hydrate.

• Re-scan: "Text-to-markup ratio" improves, and "Page weight" often improves alongside.

• Don’t pad the page with keyword text to inflate the ratio. The ratio is a proxy for substance, and stuffing is the opposite of substance.

Headings (H1, H2, H3…) are the page’s outline. Yours skip levels, repeat H1s, or use styled text instead of real heading tags, so the outline machines extract is scrambled.

Agents use the heading tree to navigate long pages and to decide which section answers a question. A broken tree means quotes get attributed to the wrong section, or sections are missed entirely.

1. Exactly one H1 per page: the page’s actual title.
2. Nest without skipping: H2 for sections, H3 for subsections. Never jump H1 to H4 for visual size.
3. If a heading looks right but is a styled <div>, make it a real <h2> and adjust CSS. Machines only see tags.

• Re-scan: "Heading hierarchy" flips to PASS.

• Don’t pick heading levels for their font size; style with CSS and keep levels semantic.

Your page HAS structured data, but it is malformed: broken JSON syntax, or types and fields that don’t exist in the schema.org vocabulary. Machines can see the block but can’t parse it.

Invalid JSON-LD is worse than it looks. The agent spends its attention budget on the block, fails to parse it, and falls back to guessing from prose anyway. You pay the cost of structured data without getting the benefit.

1. Paste the failing page into validator.schema.org. It names the exact line and field that breaks.
2. The most common breaks: a trailing comma (invalid JSON), a typo’d type name, or a price written as "$79" instead of "79" with a separate priceCurrency.
3. Fix in place, redeploy, revalidate.

• validator.schema.org reports zero errors for the page.
• Re-scan: "Structured data validates" flips to PASS.

• Don’t delete the block to make the error disappear. That trades "invalid" for "absent" and fails the presence check instead; fix it.
• Don’t hand-edit generated JSON-LD in the page. Fix the template or plugin that generates it, or the error returns on the next publish.

Time to first byte is how long your server thinks before it starts answering: the gap between an agent asking for the page and the first byte of HTML arriving. Yours crossed our threshold (under 200ms scores full marks, over 2 seconds fails).

Agents fetch with tight timeouts and many pages in parallel. A server that takes seconds to start answering gets cut off mid-queue, and a page that times out is a page that was never read. Speed here is not polish; it decides whether the content gets fetched at all.

1. Cache the HTML itself. Marketing and docs pages rarely change per visitor; serve them from a CDN or static cache so no application code runs per request.
2. If you are on serverless hosting, cold starts are a common cause: keep the function warm or pre-render the page.
3. Move slow work (database queries, third-party API calls) out of the first response; render the shell fast and fill personalised details after.

• Run: curl -o /dev/null -s -w "%{time_starttransfer}\n" https://yoursite.com and watch the number drop.
• Hit Re-scan above: "Time to first byte" improves a bucket at a time (under 500ms passes).

• Don’t answer fast with an empty shell that loads content by JavaScript. That trades this warning for the much worse "content requires JavaScript" failure, since most agents never run your scripts.
• Don’t special-case known crawler user agents onto a faster path. Serving different performance to different audiences is the pattern crawlers penalise when they detect it.

Content-Signal is a newer robots.txt directive (pushed by Cloudflare) that lets you state a granular AI policy: for example "don’t train on my content, but summarising it in answers is fine". Classic robots.txt only offers all-or-nothing per crawler; this adds the middle ground. We looked for a Content-Signal line in your robots.txt and response headers and found none.

Without a granular signal, crawlers infer your intent from blunt allow/block rules, and sites often block more than they mean to just to avoid training use. A declared signal lets you keep answer visibility (summaries, citations) while opting out of what you object to. This is an emerging standard: honoring is voluntary and adoption is early, which is why it is flagged as emerging and weighs little.

1. Decide your actual policy first: are you fine with AI training on your content? With summarisation in answers? With search indexing?
2. Express it as one line in robots.txt, for example: Content-Signal: ai-train=no, ai-summarize=yes.
3. Keep your existing User-agent rules; the signal adds nuance on top, it doesn’t replace them.

• Open yoursite.com/robots.txt and confirm the Content-Signal line is live.
• Hit Re-scan above: "Content-Signal directives" flips to PASS.

• Don’t declare signals that contradict your User-agent rules, such as ai-summarize=yes while blocking every AI crawler; conflicting instructions get you treated as unreliable.
• Don’t add the line just for the score. It is a public policy statement; say what you mean, because compliant crawlers will act on it.