The Dokeo Checks

Scans for leaked internal terms, unreleased product names, credentials, and confidential markers.

Why: Brand safety. Nothing confidential should ever reach publish.

Compares the draft against your published corpus to catch content too similar to something you already have.

Why: Duplicate pages compete with each other and dilute ranking signals.

Benchmark: Siteliner/Copyscape: pages above 50% similarity trigger Google's duplicate-content filter; 40% is the early-warning threshold used by enterprise CMS platforms.

Detects when a new piece targets the same query as an existing page.

Why: Two pages fighting for one keyword split authority and neither wins.

Benchmark: Ahrefs 2024: sites with 2+ pages on the same cluster see 38% lower average ranking for both; consolidation lifts the surviving page by a median 11 positions.

Checks heading hierarchy, section depth, and scannability.

Why: Search engines reward a clear, well-structured document.

Benchmark: Orbit Media 2024: median top-performing post is 1,427 words with 5+ H2s; HubSpot data shows posts under 300 words get 75% fewer backlinks.

Measures sentence and paragraph complexity against the level answer engines quote cleanly.

Why: If an answer engine can't parse a clean answer, it won't surface you.

Benchmark: NN/g readability research: B2B content converts best at grade 8-12; below 6 reads as dumbed-down, above 14 loses non-expert buyers.

Looks for direct-answer paragraphs, question-shaped headings, and FAQ patterns.

Why: Content shaped as answers gets pulled into ChatGPT, Perplexity, and Claude responses.

Benchmark: Zyppy 2024: pages with direct-answer paragraphs are 2.3x more likely to appear in AI overviews.

Flags statistics and strong assertions that lack a supporting citation.

Why: Unsupported claims erode the trust generative engines need to cite you.

Benchmark: Edelman Trust Barometer 2024: 64% of B2B buyers distrust content with unsourced statistics.

Checks entity clarity, attribution, and citation-worthiness.

Why: These are the signals that make an LLM name you as a source.

Benchmark: Princeton GEO study 2024: cited content sees 40% more AI-search impressions.

Subject line length (6-10 words), CTA count (exactly 1 primary), deliverability red flags (spam triggers, all-caps, exclamation abuse).

Benchmark: Marketo 2024: 6-10 word subjects peak at 21% open rate. HubSpot: emails with 1 CTA get 371% more clicks.

Headline clarity, social proof presence (testimonials/case studies), CTA above the fold.

Benchmark: Unbounce 2024: pages with social proof convert 34% higher. Spiegel Research: reviews lift conversion 270% for high-priced products.

Hook in first 8 seconds, pacing (120-160 WPM), call-to-action placement.

Benchmark: YouTube Creator Academy: 45% of viewers who watch 8s finish the video. Wistia: 120-160 WPM is the engagement sweet spot.

Discovery questions, objection handling, next-step close.

Benchmark: Gong 2024: top closers ask 11-14 discovery questions; successful calls have 4+ documented objection-handling moments.

Hook in first line, hashtag count (3-5 optimal), thread/carousel structure.

Benchmark: Hootsuite 2024: posts with 3-5 hashtags get 29% more engagement than 10+.

Hallucination indicators, confidence hedging, attribution completeness.

Benchmark: Stanford HAI 2024: 15-25% of AI-generated content contains at least one fabricated claim.

Parameter documentation, code examples, authentication section.

Benchmark: ReadMe 2024: docs with code examples see 3x higher integration success rate.

Quantified results (metrics, not vague claims), structured problem/solution format.

Benchmark: DemandGen 2024: case studies with specific metrics generate 68% more qualified leads.

Headline-to-CTA alignment, benefit-first structure, character limits.

Benchmark: Google Ads benchmarks: headlines under 30 chars see 15% higher CTR.

Scannable structure, link density, value-first opening.

Benchmark: Litmus 2024: newsletters with 3-5 links per section see peak click-through.