AI-assisted draft workflow
7 controlled stepsWe use AI to organize buyer research and approved evidence—not to manufacture experience or publish pages at scale. The draft engine checks six dimensions and 21 indicators; our separate nine-part human review decides whether the page may publish.
Article length, technical modules, comparison tables and FAQ count follow the buyer's task and the evidence available. We do not expand a page just to reach 1,500 words, 13 sections or 10 questions.
Buyer intent
Define the buyer, region, decision stage and one question the page must answer first.
Keyword & entity graph
Map real questions to one topic cluster and reject overlapping or wrong-brand routes.
Evidence map
Verify claims before the outline; unsupported facts are marked NEEDS_EVIDENCE.
Adaptive outline
Select only the modules needed to complete the buyer task—there is no fixed word count.
Draft & decision tools
Build the direct answer, checks, tables, FAQs and CTA from approved inputs.
Search presentation
Prepare metadata and structured data that match visible, verified content.
Human approval
Apply the nine-part quality gate. AI never approves or publishes its own draft.
Topic and brand alignment
12 pointsEvery page must answer one primary question and belong to a defined topic cluster. GEERPOWER content covers new machinery and turnkey production-cell projects; used-equipment inventory and inquiries are routed separately to GEEPOW Certified.
A page fails this gate when its core topic, audience or business route cannot be stated in one sentence.
Content depth and E-E-A-T
18 pointsThe draft must contribute project experience, an original decision framework, real photographs, technical evidence or practical checks beyond a generic summary. Prepared-by, reviewed-by and update information must remain visible.
Machine model, year, controller, configuration, condition, test result, price and performance are stated only when supported by reviewable records.
Search-intent coverage
12 pointsThe page identifies whether it serves informational research, commercial investigation or RFQ preparation. Direct answers resolve the initial query, while comparison, cost, risk and next-step sections support the wider decision journey.
Keywords shape the language, but the page is written to complete a buyer task rather than to repeat a query.
High-intent long-tail terms
10 pointsTitles, headings and body copy may naturally cover comparison, cost, selection, troubleshooting and quotation questions when they fit the same topic. We do not create near-identical pages for minor keyword variations.
Decision-stage language must lead to a useful action such as comparing verified evidence, preparing an RFQ or submitting product and mold data.
AI-readable answer structure
10 pointsEach guide uses one H1, descriptive H2 sections, a direct-answer block, decision checklist, FAQ, clear lists and traceable sources so search engines and AI systems can extract the answer without losing its conditions.
Structured data reflects visible content and never adds hidden ratings, testimonials, specifications or claims.
SEO structure and internal links
12 pointsThe title, description, canonical URL, heading hierarchy, image alternative text, article dates and structured data are checked before release. Internal links connect the guide to its pillar page, related decision content and the correct business route.
A new page must strengthen a topic cluster instead of becoming an isolated article.
Conversion and decision support
12 pointsCalls to action match the article context: machine comparison, project review, RFQ preparation, used-equipment due diligence or contact. FAQ content addresses realistic barriers such as scope, evidence, delivery responsibility, safety and support boundaries.
The page should help a buyer take a safer next step, not merely generate a visit.
Evergreen and reusable value
6 pointsCore guidance should remain useful beyond a short news cycle. Checklists, decision tables, direct answers, FAQ responses and evidence lists are written as reusable modules for sales training, video scripts and social content.
Dates are updated only when evidence or guidance materially changes.
Authoritative, traceable sources
8 pointsEvery publishable guide requires at least two relevant third-party sources. We prefer primary standards bodies, regulators, official technical recommendations and original manufacturer documentation over secondary summaries.
Sources are linked where they support the argument. A citation never substitutes for actual project evidence or a destination-specific compliance review.
80/100 minimum—and no unsupported claims.
A page cannot publish when its brand route is unclear, the angle duplicates an existing page, a technical or commercial claim lacks evidence, or the guide fails the source, FAQ and internal-link requirements.
- At least two traceable third-party sources
- At least two buyer FAQs and two relevant internal links
- A direct answer, decision checklist and substantive sections
- Manual review of brand, evidence, safety and commercial boundaries
External standards used by this policy
- Google Search Central — helpful, reliable, people-first content ↗
- Google Search Central — optimizing for generative AI search ↗
- Google Search Central — guidance on generative AI content ↗
- Google Search Central — Article structured data ↗
- ISO 20430:2020 — injection moulding machine safety requirements ↗
- IEC 60204-1 — electrical equipment of machines ↗
- EUROMAP — injection moulding technical recommendations ↗

