Meta tags that matter in 2026
Meta tags that matter in 2026: unique <title> (~55 chars, keyword near the front), meta description (~150 chars, a pitch not a stuffing), canonical URL, robots, viewport, and Open Graph/Twitter image at 1200×630.
Write titles for humans scanning results and descriptions for the click: promise the outcome, name the audience, skip superlatives. Duplicated titles across pages cannibalize each other.
Mistakes: keyword-stuffed titles, missing canonical on parameterized URLs (?tool=x needs one), noindex left on after staging, and 4 MB og:images that chat apps refuse to fetch.
Example: 'EAN-13 Check Digit Calculator — Free, Private | GenForge' + one canonical + 1200×630 og image = eligible, clickable, shareable.
Try: Meta Tag Generator · SERP Preview
JSON-LD without errors
JSON-LD without errors means: valid JSON (no trailing commas), the right @type with its required properties, content visible on the page (no invisible-only markup), and one script block per entity type.
FAQPage needs Question + acceptedAnswer with the full answer text; Article needs headline, datePublished and author. Validate with Google's Rich Results Test before shipping — warnings are future errors.
Mistakes: marking up content users can't see (spam flag), fake aggregateRating values (never invent ratings), unescaped </script> inside strings, and copying examples with example.com URLs left in.
Example: one FAQ pair → FAQPage with mainEntity[1]; test green, then deploy. Re-test after every template change — CMS edits break JSON silently.
Try: FAQ Schema (JSON-LD) · Article Schema
robots.txt vs sitemap: who does what
robots.txt vs sitemap: the first tells crawlers where they may go (allow/disallow per user-agent), the second tells them what exists (URLs + lastmod). One restricts, the other invites.
Put robots.txt at domain root, list the sitemap URL inside it, keep it under 500 KB, and never rely on it for secrets — disallowed URLs are still public, just unvisited politely.
Mistakes: 'Disallow: /' shipped to production (de-index everything), blocking CSS/JS (Google can't render), sitemaps with 404/canonical-mismatched URLs, and stale lastmod dates nobody updates.
Example: allow everything, disallow /admin + /*?sort=, reference sitemap.xml with 379 tool URLs and real lastmod — crawlers spend budget on pages that matter.
Try: robots.txt Generator · Sitemap Generator