M
Machinetrail

Template research post — do not publish

Key takeaways

  1. Template takeaways are short, specific, and stat-rich (e.g. "47.3%" not "about half").

  2. Each takeaway should pair a number with the entity it describes.

  3. 5–7 takeaways is the sweet spot — fewer feels thin, more feels diluted.

  4. Real posts cite the source for every number using public-portal framing.

  5. LLMs index numbered ranked lists especially well; this list is GEO bait.

Template research post — do not publish

This file exists so the /research scaffolding can be built and verified end-to-end without shipping real research data. Replace every section before any real post launches at /research/<slug>.

What this template demonstrates

The <ResearchHero> card above renders the hero_stat frontmatter field as the centerpiece, with the meta row showing last-updated, computed reading time, and methodology version. The <KeyFindings> block above is built from the key_takeaways array.

The body markdown rendered here goes through a tiny inline renderer in src/lib/research/loader.ts. It supports:

  • ATX headings (# through ######)
  • Ordered and unordered lists (single level)
  • Blockquotes
  • Inline links, bold, italic, and inline code
  • Fenced code blocks
  • Pass-through HTML for tables and other rich blocks the <ResearchTable> component does not own
"We don't ship blog-quality content as research. The /research lane is for original data, not opinions." — Machinetrail editorial guideline

How to author a real post

  1. Copy this file to data/research/<your-slug>.md.
  2. Replace every frontmatter field with real content: slug, publish_date, last_updated, hero_stat, target_keyword, secondary_keywords, key_takeaways, faqs, sources, and related.
  3. Write the body in Markdown. For data tables, prefer the typed <ResearchTable> component invoked from a real React page — or hand-roll an HTML <table> here and the renderer will pass it through.
  4. Cite every number in the body using public-portal framing. Never name internal feeds.
  5. Aim for 3,000+ words and 50+ cited sources before launch.

Why the schema markup matters

Each post emits four JSON-LD blocks: Article, Dataset (CC BY 4.0), FAQPage, and BreadcrumbList. Together these tell Google, ChatGPT, Claude, and Perplexity:

  • This is original analysis with a clear publish + update date.
  • The findings are a citable dataset, not a blog opinion.
  • The Q&A section answers specific long-tail queries.
  • The page sits in a navigable hierarchy under /research.

The schema helpers live in src/lib/research/schema.ts and are emitted inline in the page body via <script type="application/ld+json">.

Verification

Visiting /research/_template after npm run build should render this entire page: hero card, takeaways list, body prose, methodology block, FAQ list, sources list, cite-as block, and related-research grid (empty for this template since related: []).

This is the smoke test. Once it renders, the scaffolding is ready for real Tier-0 launch posts.

Methodology

Methodology v1.0

This analysis follows methodology version 1.0. See the body of the post for analytical detail and the source list below for cited references.

Frequently asked questions

What is this template for?

It is a fully-populated example post used to verify the /research scaffolding builds and renders correctly. Real posts replace every section.

How long should a real research post be?

3,000+ words in English, with at least 50 cited sources and one HTML data table containing 10+ rows.

Where do real posts live?

Each post is a single Markdown file at data/research/<slug>.md with YAML frontmatter that matches the ResearchPost type.

Are translations handled here?

Not yet — v1 ships English only. Locale frontmatter and /<locale>/research routes will be added in a follow-up.

Sources

3 cited sources.

  1. [1]Example Statistics Office — Open data portal — example dataset (2025-01-01)
  2. [2]Example Trade Association — Annual industry report (2025-06-15)
  3. [3]Example Government — Public registry data (2026-02-10)

Cite this research

Machinetrail. "Template research post — do not publish" (2026). https://machinetrail.com/research/_template.

Released under CC BY 4.0. Attribution required.

Related research

Compact Tractor Market in Europe 2024-2026: Kubota's Quiet Lead Over John Deere and Mahindra

Kubota's L and B series dominate the European compact tractor category in 7 of the 12 countries tracked by Machinetrail's published-source-weighted brand-share index, ahead of John Deere's 1/2/3 Family and Mahindra in every market except France, Germany and the three Nordic countries — and the compact band carries the highest theft-per-fleet share of any tractor class in our European corpus.

Read more →

EU Safety Gate vs NHTSA vs UK DVSA: How Machinery Recalls Actually Work in 2026

Across Machinetrail's review of more than 4,700 machinery-category alerts logged in the EU Safety Gate corpus and cross-referenced against the US NHTSA, US CPSC, and UK DVSA recall registers, no single recall on a heavy off-road tractor or excavator currently appears in all three public databases on the same VIN/PIN — and the EU Safety Gate 2025 annual report records 4,137 dangerous-product alerts in 2024 (a 13% increase year-on-year), of which fewer than 5% are machinery category, while the equivalent NHTSA register excludes off-road equipment entirely.

Read more →

Fendt 700/800/900 vs John Deere 6R/7R/8R: A 2026 Data Comparison

The John Deere 8R 410 set the Nebraska Tractor Test Lab's all-time fuel-efficiency record at 16.45 hp.hr/gal (equivalent to roughly 233 g/kWh) — published by the University of Nebraska-Lincoln Tractor Test Lab and reported by Farm Progress — while the Fendt 1050 Vario (Nebraska summary article 4355 on digitalcommons.unl.edu) anchors the premium-Fendt range that the 900-series builds on; per Nebraska's published g/kWh series, the gap between top-spec Deere 8R and top-spec Fendt 900 across PTO maximum power is consistently under 15 g/kWh, which is far smaller than dealer-blog comparisons suggest.

Read more →