E Epoch Labs

How the check works

The check is an automated, heuristic tool. It measures what can be measured objectively, and states its method and limits for every item. It isn't a ranking guarantee and isn't an audit — it's a starting point.

Check What it measures Source & limits
1. First screen Whether the first screen states what's sold, a price, and a WhatsApp/tel CTA. Fetches HTML and heuristically parses the first ~1,500 chars of visible text (not a rendered view); v1.1 will use a headless screenshot.
2. WhatsApp wa.me / api.whatsapp.com / whatsapp:// links, number format, pre-filled text. Checks the link exists and its format, not whether a message can actually be delivered.
3. Speed LCP, CLS, TBT, page weight and script count. Google PageSpeed Insights API (mobile). Marked "skipped" when there's no key or quota is exhausted — never guessed.
4. Bilingual <html lang>, reciprocal hreflang, language-switch link. A single-language site is a warn, not a fail — some SMEs are Chinese-only by design.
5. Google Business LocalBusiness/Organization JSON-LD with NAP (name/address/telephone). Checks the markup exists, not the Business Profile itself.
6. AI-crawler readiness Whether robots.txt blocks GPTBot, OAI-SearchBot, PerplexityBot, Google-Extended, ClaudeBot, etc., plus noindex. Block list follows each vendor's published docs.
7. Email trust SPF, DMARC (incl. policy), DKIM TXT records. DNS-over-HTTPS (Cloudflare 1.1.1.1). DKIM checks only 5 common selectors; not found = warn, never fail.
8. Tracking Whether GA4, GTM or Meta Pixel loads. Detects presence, not whether the setup is correct.
9. Basics HTTPS, title, meta description, viewport, canonical. Presence check.

Scoring

The overall score is a weighted 0–100, with the weights held in a single config file. We always show the score and the item-by-item results — never a score alone, because a single number explains nothing.

Honesty rules

Every result states its method and limit. Speed is "skipped" when PageSpeed quota is exhausted; an unknown DKIM selector is a warn, not a fail. We invent no benchmarks, claim no "AI-optimisation" and guarantee no rankings.