Most blog posts fail before the first paragraph is written. Not because the writing is bad, but because nobody decided what search question the article was supposed to answer. A business owner writes something they find interesting, publishes it, then asks someone to “do the SEO”. By that point, the damage is done. The title is unfocused, the URL means nothing, the headings wander, and the article half-answers three different questions instead of properly answering one. One widely quoted study found that roughly 94% of pages get no traffic from Google at all, and weak or absent optimisation is a big part of why. What changes that is learning how to optimise a blog post for SEO before you start writing.

This guide covers how to optimise a blog post for SEO and AI search, starting before you write a single sentence. It’s written for New Zealand small business owners who publish their own content. It covers the whole job: keyword, intent, structure, links, metadata, images, schema and the final review.

TL;DR: To optimise a blog post for SEO and AI search, choose one strong primary keyword before you write anything, and make sure it reflects a real question people search for. Align your SEO title, H1, URL slug, meta description, excerpt and opening paragraph around that keyword. Answer the main question early, then support it with a clear heading structure, plain language, evidence and first-hand experience. Add internal links to related pages, cite reliable external sources, include a genuine FAQ section, and use Article and FAQ schema where appropriate. Finish with a proper pre-publication review covering metadata, images, mobile layout, speed and indexability.

• Choose one primary keyword before you write anything, based on what people actually search for.
• Align your title, H1, URL, meta description and opening paragraph around that keyword.
• Answer the main question early, then support it with clear headings, plain language and real evidence.
• Add internal links, cite authoritative external sources, and include a genuine FAQ section.
• Use Article schema, validate it, and do a proper pre-publication review before you hit publish.

Choose the primary keyword before writing

The keyword comes before the title, before the URL, and before the article itself. That order matters because the keyword defines what the page is for. Once you know the phrase people actually type, everything else has a job to do: the title confirms the topic, the URL confirms it again, the opening paragraph answers it, and the headings break the answer into logical parts.

A topic and a keyword are different things. “Blogging for my plumbing business” is a topic. “How to write a blog post for a trade business” is a keyword, because it’s a phrase a real person searches. Topics are where ideas start. Keywords are where articles start.

Search intent shapes the whole piece. Someone searching “optimise a blog post for SEO” wants instructions, a clear process they can follow. A sales page or a vague opinion piece about the future of content won’t satisfy that query, no matter how well it’s written.

One page, one clear primary purpose. When a single post tries to rank for four unrelated phrases, it usually ranks for none of them. Secondary phrases still belong, but they support the main topic rather than compete with it. In this article, “blog post SEO checklist” and “optimise content for AI search” are secondary phrases. They fit naturally inside the main answer.

Be careful with keyword tools. They’re useful for comparing options, but the numbers are estimates and the suggestions are sometimes nonsense. A phrase can look attractive in a tool and be dominated in the real results by large international sites you’ll never outrank. The better habit is to search the phrase yourself and look at what actually ranks. Note what questions those pages answer, what they miss, and whether you can genuinely do better. That judgement matters even more when you use AI to help with drafting, because AI will happily write a confident article around a keyword nobody searches.

Example: how one keyword shapes the whole page

Take the keyword “optimise a blog post for SEO”. Here is how it flows through the page:

• SEO title: How to Optimise a Blog Post for SEO and AI Search
• H1: the same, or very close to it
• URL slug: /optimise-a-blog-post-for-seo-and-ai-search/
• Meta description: uses the phrase once, then describes what the reader gets
• Excerpt: a short, honest summary using the phrase naturally
• Opening paragraph: states the question and begins the answer
• H2 headings: use variations such as “keyword”, “search intent” and “heading structure” rather than repeating the full phrase
• Image filename and alt text: descriptive wording such as “checklist for optimising a blog post”, only where it genuinely describes the image
• Internal links from other pages: anchor text such as “how to optimise a blog post for SEO and AI search”

These elements should agree with each other without parroting the exact phrase in every field. Alignment builds clarity. Repetition builds nothing.

Key point: pick one primary keyword before you draft, confirm it matches a real search with real intent, and let it set the title, URL, structure and opening. Everything else in this guide depends on that decision.

Match the article to search intent

Search intent is the reason behind the search. Informational searches want an answer or instructions. Commercial investigation searches compare options before buying. Transactional searches want to buy or book now. A how-to article serves informational intent, so its job is to teach. A tutorial that pivots into a sales pitch after two paragraphs annoys the reader and fails the query.

Before writing, work out five things:

• Who is searching. Here, it is business owners and staff who publish website content.
• What they are trying to do. Publish blog posts that get found in Google and in AI tools.
• What they probably already know. That content matters, and that SEO exists. Usually not how the pieces fit together.
• What would leave their question unanswered. Vague advice like “write great content” with no practical steps.
• What useful next step they need. A checklist they can apply to their next post, and help if they want it.

Answer the main question early. Readers in a hurry get what they came for, and the page immediately shows its relevance. The detail, evidence and context follow.

Key point: know who’s searching, what they’re trying to do, and what a complete answer looks like for them. Then deliver that, without turning it into a pitch.

Build a clear heading structure

Headings establish the information hierarchy of the page. They tell readers, search engines and AI tools what subjects the article covers and how those subjects relate. Treating them as decoration wastes one of the most useful signals on the page.

Use them like this:

• One H1 for the article title, and only one.
• H2 headings for each major subject.
• H3 headings only for genuine subsections within an H2.
• Question-based headings where they reflect questions people actually search.
• Lists, tables and numbered steps where the format genuinely aids understanding. Number a process. Tabulate a comparison. Otherwise write paragraphs.

Each heading should make sense read on its own, because that is exactly how search engines and AI systems often encounter them. Clever headings that only make sense in context fail this test.

Some examples:

• Weak: “Getting started”. Improved: “Choose the primary keyword before writing”.
• Weak: “The secret sauce”. Improved: “Write for humans while making the meaning easy to extract”.
• Weak: “Wrapping up”. Improved: “Review the article before publishing”.

Key point: every heading should make sense read in isolation, because that’s exactly how search engines and AI systems often encounter them.

Give readers the short answer early

Some readers want the full detail. Many want the answer in thirty seconds. Serve both by placing a labelled TL;DR or a direct answer near the top of the article, then providing the depth below it.

A good early summary follows a few rules. It gives a genuine answer rather than a teaser. It stays accurate and consistent with the full article, because a summary that oversells or contradicts the body destroys trust. And it states important facts in complete, quotable sentences, which makes those facts easy for AI systems to lift and cite accurately.

Definitions deserve the same treatment. When you introduce a term like schema markup or search intent, define it plainly in one sentence before you build on it.

Key point: a good early answer gives genuine information, not a teaser. The rest of the article provides the evidence and detail that backs it up.

Write for humans while making the meaning easy to extract

AI search tools summarise and cite content they can interpret cleanly. The good news is that the writing habits that help machines also help people, so there’s no need to write like a robot. Do this instead:

• Use plain, precise language. Say what you mean in ordinary words.
• Name the subject explicitly. Write “the meta description” rather than “it” when a sentence stands some distance from its subject.
• Keep related material together. Finish one subject before starting the next.
• Answer one question at a time. Blended half-answers are hard for anyone to extract.
• Define terms clearly the first time they appear.
• Support claims with evidence and link to the original source.
• Add your own experience and analysis. Generic summaries are easy to reproduce and rarely worth citing.
• Separate fact from opinion. Say “in my experience” when that is what it is.
• Avoid keyword stuffing. Repeating the phrase does not strengthen the page, and it reads badly.
• Use tables for genuine comparisons and numbered lists for processes.
• Keep important information in HTML text. Facts buried only in an image, a video or a PDF are invisible to most crawlers and hard for AI tools to use.

Structure assists extraction, but the content still has to be worth citing. A beautifully formatted page of generic advice earns nothing.

Key point: write clearly for your reader first. Plain language, named subjects, one idea per paragraph. Machines can then read it too, and that’s the whole point.

Use section summaries selectively

A short takeaway at the end of a long or complex section helps readers consolidate the decision, lesson or action before moving on. Used after every small section, the same device becomes padding and readers start skipping it. Reserve summaries for the sections that carry the most weight, and make each one add something rather than restate the heading.

Two examples of useful summaries:

• After a keyword research section: “Choose the phrase by looking at the actual search results, then commit to it. The tool numbers are a starting point, and the live results page is the evidence.”
• After a schema section: “Add Article schema to every post, add FAQ schema only where visible FAQs exist, and validate before publishing. Schema describes the page. The page still has to deserve attention.”

Key point: a section summary should add something. If it only restates the heading, cut it.

Add internal links with a purpose

Internal links connect your blog post to the rest of your website. They help readers find related information, help search engines find and crawl your pages, establish the relationships between your topics, and build topical depth across the site over time. A blog post with no internal links is an orphan, and orphans are easy to ignore.

Use descriptive anchor text that says where the link goes. “Click here” tells nobody anything. Place links in context, at the point where the reader would genuinely benefit from the related page, and link only to genuinely relevant destinations.

On this site, an article about blog post optimisation naturally connects to several related pieces. If you are working on the wider site, read how to build an AI-ready website for service businesses. If you want the background on why the basics still carry the load, read why SEO fundamentals still matter in the age of AI search. If AI tools cannot find your business at all, start with why ChatGPT can’t find your business. And if you are weighing up AI-written content, the Ahrefs study on AI-generated content and SEO is worth your time.

Key point: one well-placed internal link with useful anchor text is worth more than five generic “click here” links scattered through the page.

Link to reliable external sources

Outgoing links support your claims, give readers access to the original evidence, and show that you did the research. Link to the original or authoritative source, rather than a blog that repeats someone else’s findings second-hand.

For the subjects in this guide, the primary sources are:

• Google Search Central SEO documentation, Google’s own guidance on how search works and what it expects from pages.
• Google’s introduction to structured data and the specific Article structured data guidance.
• Google’s video SEO guidance for embedded video.
• OpenAI’s crawler documentation. One practical point sits here: if you want your pages to be eligible for inclusion in ChatGPT search results, your website must allow the OAI-SearchBot crawler. Allowing it does not guarantee inclusion or citation. Blocking it guarantees exclusion.

Every external link should support a claim or provide a useful next step. Adding links to reach some imagined quota helps nobody.

Key point: link to the original source, not the article that summarised it. That’s what responsible research looks like.

Add FAQs that answer real follow-up questions

A good FAQ section answers the follow-up questions your article naturally raises. It catches related searches, resolves objections, and covers useful points that would interrupt the flow of the main narrative. It also makes question-and-answer relationships explicit on the page, which suits how AI answer systems process content.

Keep each answer concise but substantive, and keep every question and answer visible on the page. Five to eight questions is usually enough. Skip questions you can’t answer accurately, and never invent fixed numbers such as keyword densities or link quotas to make an answer sound authoritative.

Key point: FAQ sections earn their place by answering real follow-up questions completely. Teasers that point to the rest of the article are not answers.

The FAQ section for this article sits after the conclusion below.

Use the right schema markup

Schema markup, also called structured data, is code added to a page that describes the content in a format machines read reliably. In plain English, it is a label on the tin. Google explains the concept in its structured data documentation.

For a blog post, the relevant types are:

• BlogPosting or Article schema for the article itself, covering headline, author, dates and image.
• FAQPage schema where the page contains multiple visible FAQs and the implementation suits the page.
• VideoObject schema where a relevant video is embedded.
• BreadcrumbList schema where the site structure supports breadcrumbs.

Now the honest qualifications, because this is an area full of overclaiming:

• Schema helps machines understand the page. It does not directly guarantee higher rankings.
• Valid markup does not guarantee a rich result in Google.
• FAQ rich results in Google are now generally restricted to well-known, authoritative government and health websites. FAQ schema still provides clear machine-readable meaning, so it retains value, but do not add it expecting a prominent Google FAQ listing.
• Schema must describe content visitors can actually see on the page. Invisible schema is a fast way to lose trust with search engines.
• FAQPage and QAPage are different things. QAPage is for a page focused on one question where users submit answers, like a forum thread. A normal multi-question FAQ section uses FAQPage.
• Validate your structured data before publishing, using Google’s Rich Results Test or the Schema.org validator.

A short checklist for schema on a blog post: add BlogPosting schema with accurate author, dates and image. Add FAQPage schema only if visible FAQs exist. Add VideoObject only if a video is embedded. Validate, then publish.

Key point: schema is a label on the tin, not a ranking shortcut. The page still has to deserve attention.

Optimise an embedded video where applicable

Embed a video only when it genuinely supports the article. A relevant video adds real value. A filler video added “for SEO” adds load time and nothing else.

Where a video does belong, give it a descriptive title, a unique description, and a relevant thumbnail that crawlers can access. Record the upload date and duration, provide the embed URL or content URL, and add VideoObject structured data as described in Google’s video guidance. Include a transcript or substantial written coverage of the same material on the page, because the spoken content is otherwise invisible to text-based crawlers. For longer videos on YouTube, chapters help viewers and can surface key moments in search.

Key point: a video without a transcript is invisible to most crawlers. If the spoken content matters, put it on the page in text as well.

Optimise the supporting page elements

These are the elements people rush at the end. They deserve deliberate attention:

• URL slug: short, descriptive, includes the keyword. No dates, no filler words.
• SEO title: keyword near the front, roughly 50 to 60 characters, accurate.
• Meta description: around 145 to 160 characters, uses the keyword once, tells the searcher what they will get.
• WordPress excerpt: an honest summary, since themes and feeds display it.
• Featured image: relevant and high quality.
• Image filenames: descriptive, such as blog-post-seo-checklist.webp, never IMG_4021.jpg.
• Alt text: describes the image for people who cannot see it. Include the keyword only if it genuinely describes the image. Forcing it in helps nobody and reads as spam.
• Image dimensions and compression: sized for the layout and compressed, because oversized images ruin page speed.
• Author information: a named, real author with a short bio.
• Publication and modified dates: shown on the page and kept accurate.
• Canonical URL: self-referencing, to prevent duplicate content confusion.
• Mobile readability: check the actual post on an actual phone.
• Page speed: test it, then fix the worst offenders, usually images.
• Indexability: confirm the post is not blocked by robots.txt or a noindex tag.
• XML sitemap: confirm the post appears in it.

Key point: these elements get rushed because they feel like admin. They’re not. A strong meta description gets the click. A slow page loses it.

Demonstrate experience, evidence and trust

Generic summaries are easy to reproduce, which means they’re unlikely to become the best available answer to anything. AI can generate a competent generic article in seconds, and so can anyone with a free tool. What AI can’t generate is your experience.

Add first-hand experience, your own screenshots, your actual process, and observations you’ve tested yourself. Use examples that can be verified, or label a hypothetical clearly as a hypothetical. Show clear authorship, keep dates current, link to original sources, and be honest about limitations. Saying “this worked for my clients in these circumstances, and here’s where it might not apply” builds more trust than confident absolutes.

The payoff compounds. Research collected by Digidop found that a blog post generates around 60% more traffic after twelve months than in its first six. Content that deserves to be cited keeps earning long after publication.

Where a real example is needed and you don’t have one, leave a note for the author to add one later. Never invent a client story. [Christine: if you have a genuine client example of a post improving after restructuring, this section is the place for it.]

Key point: your experience is the one thing generic content can’t replicate. Use it.

Review the article before publishing

Before you hit publish, work through this checklist. Treat it as a review of the actual page, not a scoring exercise, and do not treat a green light in an SEO plugin as proof of anything.

• Keyword and intent: one primary keyword, chosen before drafting, and the article fully answers that intent.
• Title and H1: aligned with the keyword, accurate, keyword near the front.
• URL: short, descriptive, matches the topic.
• Excerpt and meta description: written deliberately, correct length, honest.
• Opening and TL;DR: the main question is answered early and accurately.
• Heading hierarchy: one H1, logical H2s, H3s only where genuine subsections exist.
• Completeness: nothing the searcher needs is missing.
• Accuracy and sources: claims checked, evidence linked to originals.
• Internal links: relevant destinations with descriptive anchors.
• External links: each one supports a claim or a next step.
• Images and alt text: descriptive filenames, honest alt text, compressed files.
• FAQs: real questions, complete visible answers.
• Schema: Article schema present, FAQ and video schema where appropriate, all validated.
• Author and dates: visible and accurate.
• Mobile layout and page speed: tested on a real device.
• Indexability and canonical: the page can be crawled, indexed and points to itself.
• Grammar and readability: read it aloud once. You will catch things a plugin never will.
• Calls to action: present, measured, relevant.
• AI-generated material: given a full human review for accuracy, tone and invented claims before publication.

After Hitting Publish

Once your post is published, your work is not over.  Go into Google Search Console and submit the post (you do have your site listed there, right?).  Post or schedule some social media posts that link to your blog post.  Add a mention of your post to your next email newsletter.

And, if relevant, make a related video that you post on YouTube, with a link to your blog post in the description, and then embed the video on your blog post page.  I wrote a checklist on video optimisation way back in 2015, and while some of the finer details have changed, the principles are still the same.

Good optimisation starts before the first sentence

Everything in this guide comes back to one decision made early. Choose a strong primary keyword, understand the intent behind it, then make the title, URL, headings, opening, content, links, metadata and schema work together around one useful answer. Do that, and the post serves all three of its audiences: the person reading it, the search engine ranking it, and the AI system deciding whether it’s worth citing. Skip that decision, and no amount of after-the-fact tinkering rescues an unfocused article.

If you’d like help choosing the right topics, structuring your content properly, or improving your existing website’s visibility in Google and AI search, get in touch about Gecko Gully’s content marketing, SEO and AEO service.  And if your wider website needs the same treatment as your blog posts, start with how to build an AI-ready website for service businesses.

Key takeaways

• Choose your primary keyword before you write a single word. The keyword defines the whole page.
• Match the article to a clear search intent. A tutorial should teach, not sell.
• Answer the main question early. Long articles that bury the answer lose readers fast.
• Plain language and clear structure serve both human readers and AI systems. Write for people first.
• Schema, metadata and supporting elements matter. They’re not optional extras.
• Your first-hand experience is your competitive advantage. Generic content is everywhere. Yours shouldn’t be.
• Do a proper pre-publication review. A green light in an SEO plugin is not the same as a post that’s ready.

Frequently asked questions

How many times should a primary keyword appear in a blog post?

Keyword density targets are outdated advice. Use the exact phrase in the title, URL, meta description and opening, then write naturally using variations and related terms. If the phrase feels forced anywhere, remove it.

Does every blog post need a TL;DR?

A TL;DR earns its place on longer or more complex articles where readers benefit from an upfront answer. A short post that answers its question in the first paragraph already does the job.

Does adding FAQ schema improve Google rankings?

Valid markup doesn’t guarantee a rich result, and Google now generally restricts FAQ rich results to well-known government and health websites. FAQ schema still gives machines a clear, structured view of your questions and answers, so it retains value beyond Google’s FAQ display.

How long should an SEO blog post be?

Long enough to answer the search query completely, and no longer. A thin page fails because it leaves questions unanswered. A padded page fails because readers give up. Word count targets are a distraction.

Can AI-written blog posts rank in Google?

Google evaluates quality rather than authorship. In practice, unedited AI content tends to be generic, and generic content rarely becomes the best answer to anything. Use AI for drafting if you like, then add your own experience, verify every claim, and review the result properly before publishing.

What is the difference between SEO and AI search optimisation?

SEO helps search engines crawl, understand and rank your pages. AI search optimisation helps AI answer systems identify, extract, summarise and cite your information. Most of the second builds directly on the first: clear structure, direct answers, credible evidence and technically accessible pages serve both. One practical addition for AI search is allowing crawlers such as OAI-SearchBot, which is required for eligibility in ChatGPT search results.

How soon should an old blog post be updated?

Review it whenever the facts, tools or guidance it describes change, and check your important posts at least once a year. Update the content genuinely and correct anything outdated. Adjust the modified date. Changing the date without changing the content fools nobody.

Do I need to optimise every page element, or just the content?

All of them work together. Strong content with a weak meta description gets fewer clicks from search results. Strong metadata on a thin page earns the click but loses the reader. The URL, title, meta description, images, schema and the content itself each play a role. Doing most of them well is fine. Ignoring one category consistently will cost you.