RESEARCH · TECHNICAL

How this site is built, and why it is built that way

Every marketing page on rankedtag.com is pre-rendered to static HTML at build time, so the full content is in the served document before any JavaScript runs. The blog is server-rendered from a Postgres database on request. One organisation entity is injected into every page, and a verification script blocks the build if any page is thin, orphaned, broken or carries invalid structured data.

Last updated . Written and verified by Bhushan Raj Shakya, founder of RankedTag.

KEY FACTS

Rendering
Static pre-rendering at build time for marketing pages; server-rendered HTML for blog posts
Why
Assistant crawlers generally do not execute JavaScript, so content behind it is invisible to them
Entity
One Organization and WebSite graph injected into every page at build time
Crawlers
Assistant agents explicitly allowed in robots.txt by documented name
Quality gate
A script that fails the build on dead links, thin pages, orphans or invalid JSON-LD
Hosting
Static output on a CDN, with a small number of serverless endpoints for the tools and the blog

Why the rendering decision is the one that matters

A single-page application serves a near-empty HTML document and builds the page in the browser. Google documents that it processes JavaScript in a separate rendering stage, which means content behind JavaScript is seen later and sometimes not at all Google Search Central, JavaScript SEO basics. Assistant crawlers are less forgiving again: their documentation describes fetching pages, not executing applications OpenAI, bots and crawler documentation.

The practical consequence is that a marketing site built as a client-rendered application can look complete in a browser and arrive as an empty shell to the systems that decide whether to cite it. Every page here is therefore written to static HTML at build time, with the full text, headings and structured data present in the served document. The JavaScript that hydrates the page afterwards adds interaction, not content.

The blog is different, because its content lives in a database that changes without a rebuild. Those pages are rendered to HTML on the server at request time, which produces the same outcome by a different route: a crawler receives finished HTML either way.

One entity, injected everywhere

The Organization and WebSite nodes are defined once and injected into every page at build time, and the same definition is imported by the server-rendered blog. One definition means the static pages and the blog cannot drift apart, which is the usual way an entity graph quietly becomes inconsistent.

Page-level schema references those nodes by identifier rather than repeating them, which is the schema.org idiom for "the same entity" and avoids publishing three slightly different versions of the same organisation Google Search Central, intro to structured data markup. The founder is one Person node with one identifier, referenced by the organisation, by the about page and by every blog byline.

Structured data is only published where it reflects what a reader can see on the page. No aggregate ratings are published anywhere on this site, because there is no genuine review data behind them, and Google requires that markup represent visible content Google Search Central, FAQ structured data.

What the robots file says, and why

Assistant crawlers are allowed explicitly, by the names their operators document, rather than being left to a default. OpenAI documents OAI-SearchBot, GPTBot and ChatGPT-User with distinct purposes OpenAI, bots and crawler documentation. Anthropic documents ClaudeBot, Claude-User and Claude-SearchBot Anthropic Support, does Anthropic crawl data from the web. Perplexity documents PerplexityBot and Perplexity-User Perplexity, PerplexityBot documentation. Google-Extended is a control token with no user agent of its own, governing whether crawled content may be used for training and grounding in Gemini apps without affecting Google Search Google Search Central, Google's common crawlers.

Allowing them is a deliberate commercial choice rather than a default. The content on this site exists to be read and cited; a citation with attribution is the outcome being optimised for, which is also why the terms explicitly permit quotation with attribution.

The quality gate that blocks the build

A verification script runs against the built output and fails the build rather than warning. It checks every internal link for dead ends and redirect chains, every page for the minimum length, every page for reachability within two clicks of the homepage, every structured data block for whether it parses, the sitemap against the built pages in both directions, heading order, and whether any page uses a styling class whose stylesheet it never imports.

That last check exists because of a real failure: a stylesheet defining the prose layout was not imported by eight pages, which shipped as unstyled full-width text. The check now catches it before a human has to notice. The gate has also caught ten orphaned pages and a page whose review date rendered as plain body text, which is the argument for having it at all.

Frequently asked questions

Do AI crawlers execute JavaScript?

Generally no. Operator documentation describes fetching pages rather than rendering applications, so content that only appears after JavaScript runs is at risk of being invisible to them.

Is server-side rendering required for AI visibility?

Not strictly, but the content has to be in the HTML a crawler receives. Static pre-rendering and server-side rendering are two ways of achieving the same thing.

Does blocking GPTBot remove a site from ChatGPT?

Not exactly. OpenAI documents separate agents for training, search indexing and user-triggered fetches, so blocking one affects one use. The names and purposes are in OpenAI’s own documentation OpenAI, bots and crawler documentation.

Why publish no aggregate ratings?

Because there is no genuine review dataset behind them. Structured data has to reflect visible content, and inventing a rating to earn a star in a results page is the kind of thing that earns a manual action instead.

Can I see the verification script?

It is not published as a product, but the categories it checks are listed above and the founder review applies the same checks to an applicant’s domain.

Official resources

FOUR SLOTS A MONTH

The site is the proof of the method.
Let it look at yours.

The 52-check founder review is run personally and comes back within 48 hours, with a real opinion even when the answer is no.