---
name: rankizz-write-human-blog-post
description: Research, outline, draft, edit, or rewrite a source-backed blog post that sounds natural and useful rather than formulaic or AI-generated. Use when a user asks for a blog article, long-form educational post, thought-leadership draft, humanized content, editorial rewrite, or complete SEO content package with title, description, headings, links, image briefs, and limitations. Rankizz keyword and SERP data is optional; web research and honest source handling are required for current factual claims.
---

# Write a human blog post

Write for a defined reader with a real problem. Produce original value, not a rearranged search-results summary. Never fabricate experience, interviews, quotations, data, or product use.

Read [references/research-and-human-writing.md](references/research-and-human-writing.md) before researching or drafting.

## Establish the brief

Resolve:

- intended reader and what they already know
- problem, decision, or task the post should help with
- publisher, subject-matter expertise, and business relevance
- primary topic and search intent
- desired action after reading
- region, language, and publication date
- author voice sample, if available
- facts or opinions the author can uniquely contribute
- required length only when the publisher has a real constraint

Do not ask for a target word count as an SEO ritual. The post is done when it answers the reader's question without filler.

## Research before outlining

Browse for every current, disputed, quoted, regulated, medical, legal, financial, pricing, product, or statistical claim. Prefer sources in this order:

1. laws, standards, official documentation, datasets, filings, and original research
2. the subject's own product pages, policies, changelog, and help center
3. named experts and reputable reporting with transparent sourcing
4. secondary commentary only when it adds analysis or lived experience

Build a compact evidence ledger:

| Claim | Source | Published or updated | Direct evidence or inference | Allowed wording |
| --- | --- | --- | --- | --- |

Use direct quotes sparingly and accurately. Link close to the supported claim. If reliable evidence is unavailable, remove the claim or label the uncertainty. Never turn a search snippet into evidence.

Rankizz is optional. Use keyword, SERP, Search Console, or analytics tools only when connected data will change the angle, outline, or prioritization. If a Rankizz call is unavailable because of entitlement or credit balance, tell the user: "This feature is unavailable because the connected account does not have the required entitlement or credit balance." A writing task can proceed from local and public evidence without Rankizz.

## Find an original contribution

Before drafting, state the post's contribution in one sentence. It should contain at least one of:

- first-hand experience supplied by the author
- original data, calculation, test, or example
- a useful synthesis that resolves conflicting advice
- a decision framework tied to a specific audience
- a concrete implementation, teardown, or case
- a defensible point of view with evidence

Stop if the only plan is to paraphrase pages that already rank. Ask for expertise or narrow the topic until the post can add value.

## Build the outline

Match the structure to the reader's task, not a fixed blog formula.

1. Answer or orient the reader in the opening.
2. Put prerequisites and definitions only where needed.
3. Order sections by decision sequence, workflow, chronology, or importance.
4. Give each heading one distinct job.
5. Place examples, evidence, and caveats beside the claim they support.
6. End with the next useful action or a specific unresolved implication.

Use one H1. Use sentence case headings. Avoid a generic `Introduction`, `Benefits`, `Challenges`, or `Conclusion` section unless the content genuinely calls for it.

## Draft with a human voice

If the user provides a voice sample, match its sentence length, vocabulary, paragraph openings, punctuation, transitions, and degree of formality. Preserve idiosyncrasies that help the writing sound like that author.

Without a sample:

- prefer plain, specific words and active voice
- vary sentence and paragraph length naturally
- use concrete nouns, named actors, dates, numbers, and examples
- state opinions only when the author can own them
- admit uncertainty where the evidence is mixed
- use first person only for experience the author actually supplied
- let the topic determine warmth, humor, or restraint

Human writing is not fake messiness. Do not add invented anecdotes, random slang, grammar mistakes, theatrical asides, or unsupported emotion.

## Remove common AI patterns

Delete or rewrite:

- `Let's dive in`, `Here's what you need to know`, and other announcements
- inflated claims about a changing landscape, pivotal moments, or transformations
- empty adjectives such as `robust`, `seamless`, `vibrant`, and `game-changing`
- vague attribution to experts, studies, users, or reports
- repetitive groups of three
- strings of same-length sentences or manufactured one-line punchlines
- `not just X, but Y` constructions used as decoration
- paragraphs that merely restate their heading
- generic upbeat conclusions
- repeated bold labels and overformatted lists
- em dashes and en dashes

Do not remove a phrase simply because it appears on an AI-pattern list. Look for clusters, preserve correct technical language, and keep the author's real voice.

## Add the publication package

After the draft, provide:

- suggested URL slug
- SEO title and meta description that follow the publisher's rules
- one H1 and logical H2/H3 outline
- internal-link opportunities with descriptive anchor text
- external source links
- hero and inline image briefs with purpose-specific alt text
- author/byline and review needs
- `Article` or `BlogPosting` structured-data fields only when visible page data supports them
- a fact-check date for time-sensitive posts

Do not stuff the target keyword into the title, headings, links, and every image. Search optimization supports the article; it does not replace usefulness.

## Do and do not

| Do | Do not |
| --- | --- |
| Write to one identifiable reader | Write for "everyone interested in the topic" |
| Start with the answer, tension, or concrete situation | Open with a dictionary definition or sweeping history |
| Name sources and link claims | Say "experts agree" without a source |
| Use a real example with relevant detail | Invent a customer, quote, test, or personal story |
| Keep uncertainty visible | Turn an inference into a fact |
| Vary rhythm in service of meaning | Randomly alternate sentence lengths to mimic humanity |
| Use the author's actual point of view | Manufacture hot takes or false vulnerability |
| Cut filler regardless of target word count | Pad the post because long content is assumed to rank |
| Disclose affiliate or commercial relationships | Hide compensation below the article or behind vague wording |
| Update the substance before changing a date | Refresh a timestamp to make stale content look new |

## Examples

### Weak opening

> In today's rapidly evolving digital landscape, project management has become more crucial than ever. In this comprehensive guide, we will delve into everything you need to know.

It names no reader, problem, evidence, or point of view.

### Better opening

> A ten-person agency rarely needs a project-management platform with twelve permission levels. It needs a place where clients can find the latest file and the team can see who owes the next decision. This guide compares the workflows that matter at that size.

This opening defines the audience, rejects an assumption, and previews the article's decision criteria.

### Unsupported experience

Bad:

> After testing the tool for three months, I found its reports unreliable.

Use only if the author actually performed and documented that test. Otherwise write:

> The documentation exports reports as CSV, but the available evidence does not establish how reliable the reports are with large accounts.

### Natural transition

Weak:

> Additionally, it is important to note that pricing is another crucial factor to consider.

Better:

> Price changes the recommendation. The starter plan covers two editors; the next tier charges for five whether you need them or not.

## Limitations

- A model cannot supply first-hand experience that the author did not provide.
- Web research cannot confirm private workflows, unpublished pricing, or product performance behind inaccessible accounts.
- A natural voice does not guarantee originality, accuracy, rankings, conversions, or legal compliance.
- High-stakes topics require qualified human review.
- Do not publish direct quotations, licensed images, or substantial copyrighted passages without permission and appropriate limits.
- If browsing is unavailable or forbidden, label current claims as unverified and do not invent them.

## Output template

```markdown
# Content brief
- Reader:
- Reader's job/problem:
- Primary intent:
- Business relevance:
- Original contribution:
- Voice:
- Desired action:

## Evidence ledger
| Claim | Source | Date | Evidence level | Notes |
| --- | --- | --- | --- | --- |

## Outline
H1:
  H2:
    H3:

## Draft
# [Specific H1]
[Opening that answers or frames the real problem]

## [First decision or task section]
[Claim, evidence, explanation, example, limitation]

## [Next useful section]
[Continue only as long as the reader needs]

## [Specific closing heading when useful]
[Concrete next action or honest conclusion]

## Publication package
- Slug:
- SEO title:
- Meta description:
- Internal links:
- External sources:
- Image briefs and alt text:
- Author/reviewer:
- Structured data fields:
- Facts to recheck before publishing:

## Editorial QA
- [ ] No invented experience, quote, source, statistic, or result
- [ ] Every current factual claim is sourced
- [ ] Draft adds original value beyond source summaries
- [ ] Voice matches the author and audience
- [ ] AI-pattern and repetition pass completed
- [ ] Commercial relationships are disclosed
```