Blog Writing: How to Write One Post, Start to Finish

The pass that separates a draft from a post. It is the job, not the tidy-up.

Nearly every page ranking for this phrase gives you the same list. Pick a topic. Write a catchy headline. Add images. Proofread. None of that is wrong — it is just not where the time goes.

The time goes into decisions: which question the post answers, what you may assume the reader knows, which sentence opens it, which heading earns a skim, and what you delete on the second pass. This is the post-level chapter of our content marketing strategy guide, which decides whether a post is worth writing at all.

One boundary, stated once. Content writing is the wider discipline — web copy, email, social, product pages, long-form. This page is narrower on purpose: one blog post, from blank document to published URL.

What blog writing actually involves

Blog writing is a sequence of nine decisions, not a burst of typing: choose the question, place the reader, write the opening, build the structure, write the headings, control the paragraphs, self-edit, handle the query, and give the reader somewhere to go.

Three things sit outside this page, and ignoring that is how a post becomes a hub nobody finishes:

• What to write, and why. Topic selection, clusters and business fit live in our content marketing guide.
• The brief. If someone else set the angle, audience, query and sections, that document is a brief — see our content brief guide.
• The post's afterlife. Turning one post into a carousel, a newsletter and a script is content repurposing.

Decide the one question your post answers

A good post answers one question completely. Write that question down, in the reader's words, before any of the body — and if you cannot answer it in one sentence, you do not understand it well enough to write 2,000 words about it.

The test that you have only one question: write the 40-to-60-word answer. If it needs the words "and also", you have two posts. Split them. Two half-answered questions rank for neither.

Check what the question implies, too. "How to" wants steps. "What is" wants a definition and a boundary. A comparison wants a table and a recommendation. That match is search intent, and getting it wrong makes every later decision harder.

Start where the reader already is

The single biggest waste in blog writing is the first 300 words spent re-establishing what the reader already knows. Someone who typed "how to write a blog post" does not need a blog defined for them. Start at their actual gap.

Two cheap ways to find where they are:

• Read the questions you already get. Sales calls, support tickets, comments, the WhatsApp group. Real questions come pre-loaded with what the asker already knew.
• Read the top three competing pages for what they assume. Not to copy them — to find the floor. If all three define the term in paragraph one, you can start above it.

Then write one line at the top of your draft: "This reader already knows ___ and is stuck on ___." Delete anything that argues with the first blank.

Write an opening that earns the scroll

An opening has one job: make the next paragraph feel necessary. It is not a warm-up, not a summary, and not a place to establish that the topic matters. If a sentence would survive being pasted into an article on a different subject, cut it.

Three openings that work:

1. The correction. Name what most people get wrong, flatly. "Most drafts are too long because nobody removed the first paragraph."
2. The specific scene. One situation the reader recognises. Specificity buys trust that adjectives cannot.
3. The direct answer. Answer in the first two lines, then spend the post earning it. Search snippets and AI answers both lift self-contained answers.

Openings that do not work: "in today's fast-paced world", "we all know that content is king", a rhetorical question aimed at the reader, or a definition of a word they typed into a search box. Delete the sentence explaining what the post will explain.

Structure a post to be skimmable and quotable at once

Two readers hit every post. One skims — headings, bold, lists, out. One quotes — a search engine, an AI answer, or a colleague pasting a line into Slack. Both are served by the same move: open every section with a self-contained answer, then add the detail.

Four habits that serve both:

• One idea per H2. Two answers means two sections. No answer means it is a paragraph.
• Answer in the first two sentences under each heading, with no pronoun pointing back at the heading.
• A list, table or example every 150 to 250 words. Unbroken prose is where phone readers leave.
• Order sections by the reader's sequence, not yours. Which is why the question comes before the outline.

Write headings that are answers, not labels

A label tells the reader what the section is about. An answer tells them what it concludes. Answers win, because a skimmer who reads only headings still leaves with the argument — and because headings phrased as real sub-questions match how people search and how AI systems pull passages.

Two limits: keep a heading a heading, not a sentence, and front-load it — long headings wrap to three lines on a phone.

Paragraph discipline that survives a phone screen

Write for a 390-pixel-wide screen, because that is where most of your readers are. A paragraph that looks tidy on a laptop becomes eleven lines on a phone, and eleven lines reads as a wall. One idea per paragraph, one or two sentences, and no shame in a single line.

• Cap most sentences at around 25 words. Longer is allowed when the length is doing something. It usually is not.
• Front-load the sentence. "Cut the first paragraph if it only warms up" beats "If it only warms up, the first paragraph should be cut".
• Check the draft at phone width. Narrow the window and you will find a paragraph to split and a heading to shorten.

The self-edit pass: run this checklist on your own draft

Most drafts are one editing pass away from being good, and that pass is subtractive: a draft that survives all nine checks comes out shorter than it went in. Do it in a separate sitting from the writing, top to bottom, in this order. The first check is the one people refuse.

1. Cut the first paragraph. Read the draft from paragraph two. If it still works — and it nearly always does — paragraph one was you clearing your throat.
2. Delete every sentence that only restates the previous one. Restatement feels like emphasis while writing and reads as padding afterwards. One idea, once.
3. Replace every abstraction with the specific thing you meant. "Optimise your content" means nothing; "cut the first paragraph and rewrite the H2s as questions" means something. If you cannot name it, you have not finished thinking.
4. Delete the hedges. "Arguably", "it could be said", "generally speaking", "in many cases". Say it or remove the claim. Hedging is the tell of a writer who does not know.
5. Read it aloud. Every sentence you stumble over is one a reader stumbles over. Fix the stumble, not the grammar.
6. Check each heading against its section. If the section drifted, one of the two is wrong — usually the section has two ideas in it.
7. Verify every number, name, date and link. Remove anything you cannot source. One wrong figure costs more trust than the figure bought.
8. Re-read the opening last. Now you know what the post says, you can write the sentence that promises it.
9. Read it on a phone. Last check, no exceptions.

Where AI helps in blog writing, and where it goes hollow

AI is genuinely useful either side of the drafting and weakest in the middle of it. It is good at collating research you then verify, at proposing structures so you can reject two, and at a second opinion on whether a section answers its heading. It is poor at the sentences themselves, which is where readers decide whether to stay.

That last row is our opinion, from reading a lot of AI-assisted drafts, not a Google position. Readers do not bounce because a machine helped. They bounce because nothing in the paragraph could only have been written by someone who had done the work.

What Google actually says. Its guidance on creating helpful content states that what matters is how the content was created and why, and that "That can include automated, AI-generated, and AI-assisted content" (Google Search Central, last updated 10 December 2025). The line that bites is about intent: "If you use automation, including AI-generation, to produce content for the primary purpose of manipulating search rankings, that's a violation of our spam policies."

Its separate page on AI-generated content warns that "using generative AI tools or other similar tools to generate many pages without adding value for users may violate Google's spam policy on scaled content abuse", and defines that abuse as "when many pages are generated for the primary purpose of manipulating search rankings and not helping users" (Google Search Central, 10 December 2025; spam policies, 28 August 2026). On telling readers, it suggests you "consider adding information on how your content was created in a way that makes sense for your audience".

So the rule is not "no AI". It is that the post has to be worth reading, and volume is not a strategy. Publishing to that standard weekly is a resourcing decision more than a writing one, and it is the work we take on in content marketing.

The SEO part, without turning this into an SEO post

Four things, and then stop. One target query per post, headings that mirror the real sub-questions around it, links out to the pages that go deeper, and a title that describes the page honestly. The rest of the on-page checklist matters less than whether the post answers its question.

• One query, one post. Two posts aimed at the same query compete with each other and split whatever authority you have.
• Headings as sub-questions. The same discipline as writing headings that answer, so the SEO benefit is free.
• Link out deliberately. Two to four links to the pages carrying the next step, with anchors that describe the destination — the architecture side is internal linking.
• Word count is not a ranking factor. Google's own guidance asks: "Are you writing to a particular word count because you've heard or read that Google has a preferred word count? (No, we don't.)" (Google Search Central, last updated 10 December 2025). Write until the question is answered, then stop.

Expect patience. A new post usually takes three to six months to settle into a position that tells you anything, and longer on a competitive head term. Judge its first fortnight on whether readers finish it, not on where it ranks.

End by giving the reader somewhere to go

An ending is not a summary. The reader just read the post, so repeating it wastes the moment they are most willing to act. Give them one next move and one link that serves it.

Three endings that work, and one that does not:

• The next decision. Name what they must choose next, and link the page that helps them choose.
• The smallest first action. "Open your last post and cut the first paragraph" beats "start creating better content today".
• The honest limit. Say what the post does not cover and where that lives. Readers trust a page that admits its edges.
• The one to avoid: "In conclusion, blog writing is an essential part of any marketing strategy." Nothing happens next.

Your next move: take the last post you published, run the nine-step pass on it, and count what you delete. That number tells you more than any tool will.