Research

Research: competitors and Next.js metadata issues (October 2026)

Checked 2026-10-14 (AEO sections added the same day after the SEO/AEO review). Weekly downloads from https://api.npmjs.org/downloads/point/last-week/<pkg> for 2026-10-06 → 2026-10-12. Release dates from the npm registry. Every link was opened during this review.

Download numbers

Package	Downloads/week	Latest	Notes
schema-dts	3,138,833	2.1.0 (2026-10-02)	Types only; depends on schema-dts-lib
@vercel/og	2,142,440	—	Same engine as next/og
next-sitemap	638,748	4.2.3 (2023-09-06)	4 runtime deps
next-seo	472,622	7.3.0 (2026-07-29)	No runtime deps; peers next >=13.4, react >=18.2
react-schemaorg	138,712	—	JSON-LD component typed with schema-dts

Competitors

next-seo

• Active: 7.1.0 (Jan 2026), 7.2.0 (Feb 2026), 7.3.0 (Jul 2026); ~8.5k stars; 0 open issues at time of check.
• v7 README: "For standard meta tags (<title>, <meta>), use Next.js's built-in generateMetadata". The App Router value is JSON-LD components; Pages Router lives under next-seo/pages (README).
• No sitemap, robots, hreflang checks or CI tool. → next-herald complements rather than replaces it for JSON-LD; migration guide covers both.

next-sitemap

• Last release 4.2.3 on 2023-09-06; repo last pushed 2026-05; 2 open issues.
• Postbuild script writing to public/. With Next 16 on Vercel those files are not deployed: iamvishnusankar/next-sitemap#1048. Also #1050 (duplicate Content-Type).
• Runtime deps: @corex/deepmerge, @next/env@^13.4.3, fast-glob, minimist.
• → Strongest gap: a maintained App Router-native sitemap index + robots solution. ADR 0001/0007.

Native Next.js 16 (latest 16.4.0, canary 16.5.0-canary.10)

Supports metadata/generateMetadata, sitemap.ts with images, videos, alternates.languages and generateSitemaps (docs, generateSitemaps), robots.ts, opengraph-image, ImageResponse. Gaps: no sitemap index, no news extension, shallow metadata merging, no JSON-LD helper beyond a guide that warns JSON.stringify is not XSS-safe, no checker.

Open upstream issues (all open on 2026-10-14):

Issue	Summary	Impact on next-herald
vercel/next.js#98684	Clicking a <Link> while its prefetch is in flight commits with no <title> (pending segment-cache entry)	Client-side only; documented; not detectable by herald check
#97472	Soft navigation leaves previous route's OG/Twitter tags mounted (16.3, Vercel)	Documented
#95836	not-found.tsx metadata replaced after hydration	Documented
#83510	Duplicate metadata after server-action revalidation	Documented
#80583	Link prefetch calls generateMetadata when streaming	Docs: cache metadata data ("use cache")
#95406, #96594	htmlLimitedBots customisation breaks PPR resume / caching and drops metadata	herald check --ua compares bot and browser HTML
#98126	Per-param sitemaps under a dynamic segment collapsed (16.3 regression)	ADR 0007 avoids that layout
#87314	sitemap.ts and sitemap.xml/route.ts conflict	Large-site mode has no sitemap.ts
#99055	revalidate sitemap emitted as static in adapter outputs	Documented
#99715	Metadata routes send max-age=0 instead of s-maxage	Documented
#98007	Route handlers (incl. sitemaps) not compressed by next start	Gzip option for the index handler considered (v0.3)
#54665	sitemap.ts with custom pageExtensions	herald routes reads pageExtensions from config

schema-dts and others

• schema-dts: complete schema.org types, no runtime, no serializer. We stay type-compatible (ADR 0005).
• react-schemaorg: small JSON-LD component on schema-dts types.
• Validators: Rich Results Test, Schema.org validator — web tools, not CI. Lighthouse SEO audits run a browser per page.

Beyond Next.js: Nuxt SEO, Astro, WordPress (added 2026-10-14)

Full matrix and sources: SEO/AEO review.

• Nuxt SEO is the best-in-class benchmark (@nuxtjs/seo 124,014/wk; @nuxtjs/sitemap 306,316; @nuxtjs/robots 259,772; nuxt-schema-org 175,745; nuxt-og-image 182,419; nuxt-llms 43,857): sitemap, robots with AI directives (Content-Signal and Content-Usage), schema.org graph with @id linking, link checker, nuxt-llms. next-herald aims at Nuxt SEO parity for Next.js.
• Astro: @astrojs/sitemap (3,814,836/wk) with sitemap index and i18n; meta via community packages (astro-seo 103,492/wk).
• WordPress: Yoast SEO Premium has Bot Blocker toggles (GPTBot, CCBot, Google-Extended) and IndexNow; Rank Math has IndexNow in the free plugin; both generate llms.txt (Yoast, Rank Math).
• Lighthouse (5,404,521/wk) and unlighthouse (40,625/wk) run SEO audits but need Chrome; herald check covers equivalents over HTTP (#47, #48).

AEO stance

Summary; details in AEO stance and ADR 0012.

• Works (documented by engines): crawlable server HTML; AI search bots allowed (OpenAI: OAI-SearchBot opt-out means not "shown in ChatGPT search answers"); normal Google Search eligibility for AI Overviews/AI Mode; accurate dates; entity markup that matches visible content; IndexNow for Bing and other IndexNow engines.
• Unproven (experimental opt-ins, v0.5): llms.txt / llms-full.txt, markdown alternates.
• Hype: see below.

Positioning derived from this

A Next 16 App Router–only kit that returns native objects, fills the native gaps (index, news, deep merge, safe JSON-LD, hreflang), and ships a dependency-free CI checker (a headless-free Lighthouse-SEO alternative for CI). It keeps sites legible to AI search with purpose-split crawler rules, an entity graph and no-JS visibility checks. Not: "faster", "better rankings", "gets you cited by AI", or "replaces Next metadata".

Rejected ideas

Idea	Why rejected
<Seo> head component (next-seo style)	Conflicts with Next's head management and streaming metadata (ADR 0001)
Postbuild script writing to public/	Not deployed on Next 16/Vercel (next-sitemap#1048); outside build output
Reading .next manifests for routes/check	Internal, unstable (ADR 0008/0009)
Deriving siteUrl from Host	Breaks Cache Components, enables host-header poisoning (ADR 0006)
Headless browser crawling (Puppeteer/Playwright) in herald check	Heavy dependency, slow; ADR 0002
schema-dts dependency / full generated vocabulary	ADR 0002/0005; curated + escape hatch
Bundled fonts and hosted OG service	Package size, cost (ADR 0011)
AI-generated titles/descriptions	Out of scope; quality and cost concerns; not a technical-SEO check
Auto lastModified = build time	Misleads crawlers; Google uses lastmod only if consistently accurate
Patching Next router bugs (e.g. #98684) client-side	Fragile, fights the framework; report upstream instead
Pages Router / Next 15 support	ADR 0003
Keyword density / content scoring rules	Not reliable signals; noise in CI

AEO: rejected and hype

From the SEO/AEO review, verified 2026-10-14.

Idea	Verdict	Reason
llms.txt as a ranking/citation lever	Hype (shipped only as an experimental opt-in, #55)	A proposal (llmstxt.org) with no standards body; no major AI search engine documents using it. Google's John Mueller: "AFAIK none of the AI services have said they're using LLMs.TXT … it's comparable to the keywords meta tag" (SEJ). Google: "You don't need to create new machine readable files, AI text files, or markup" (AI features).
"GEO scoring" / AI-visibility scores	Reject	Not measurable from HTML; same reason as keyword density.
speakable schema	Reject	Beta, limited to news publishers in English in the US, aimed at Google Assistant (speakable).
HowTo builder	Reject	HowTo rich results removed (2023).
Marketing FAQPage as a rich-result win	Reject	FAQ rich results limited to well-known government and health sites since 2023; builder kept without the promise.
Google Indexing API for ordinary pages	Reject	Officially limited to JobPosting and BroadcastEvent in VideoObject (quickstart); needs a service-account credential.
Blocking Google-Extended to "stay out of AI Overviews"	Reject as advice	AI Overviews use Googlebot (AI features).
Accept-header negotiation in library code	Reject	Reads request headers (ADR 0006). proxy.ts recipe only (AEO stance).
AI-generated FAQ blocks / "answer-first" rewriting	Reject	Content generation is out of scope; risks scaled-content spam policies.
Pay-per-crawl integration	Reject (out of scope)	CDN-level feature, not something a framework library can enforce.
robots.txt as enforcement against bots	Reject the claim	Compliance is voluntary; use WAF/bot management.
Headless rendering to "see what AI sees"	Reject	Most AI crawlers do not render JS, so raw server HTML is what they see (ADR 0008).
A single aiCrawlers toggle	Replaced	Merged training, search and user-fetch bots; blocking "AI" wholesale would also remove a site from AI search answers (ADR 0012).