Most checklists for this mix three separate jobs together. Deciding what to write, writing it, and preparing it for publication are different tasks done at different moments, and a single list of eight items forces you to jump between them.
So this is twenty four checks in three groups, in the order the work happens. Half of them happen before a word is written, which is where most of the ranking is decided.
How this checklist was put together
The pages ranking for this term run between 2,400 and 3,000 words with eight to thirteen items. They cover the on-page mechanics well. Two things are missing from all of them: none has a section on writing for AI assistants, which now answer a large share of these queries directly, and none lists the practices that used to be advice and now do harm. Both are here.
Before you write: 9 checks
This is the half that decides whether the page can rank at all. Skipping it means writing something excellent that competes for a term nobody searches, or one you cannot win.
1. Confirm the term is real. It has search volume and it is a phrase people actually type. Terms invented in a brainstorm are the most common reason a good page gets no traffic.
2. Check the intent by looking at the SERP. Search it and see what ranks. If the first page is all product pages and yours is a guide, you will not rank no matter how good the guide is. Match the format that is already winning.
3. Check you are not already covering it. Search your own site for the term. If an existing page targets it, improve that page instead. Two pages competing for one term is the most common self-inflicted SEO problem.
4. Read the top five results properly. Not to copy them. To find what they all include, which tells you what the searcher expects, and what none of them includes, which is your reason to exist.
5. Decide your differentiator before writing. Original data, a tool, real experience, a table nobody else has. If you cannot name it, you are about to write the sixth version of the same page. The tools that help draft against that differentiator are compared in our roundup of the best AI copywriting tools.
6. Write the outline as questions. Every heading answering a question someone would ask. This makes the structure useful for readers and quotable for assistants.
7. Pick one primary term and three or four related ones. Related terms are the ones the top results all mention. They are not synonyms to sprinkle, they are subtopics you must cover.
8. Decide the format. Guide, listicle, comparison, how-to. Established in check 2, written down here so it does not drift while writing.
9. Choose the URL slug. Short, readable, containing the term, no dates and no numbers you will want to change. This is the one element that is expensive to fix later.
While you write: 9 checks
10. Answer the question in the first hundred words. Not after a story about how important the topic is. Both readers and assistants take the answer from near the top, and burying it costs you both.
11. Use the term naturally in the first paragraph. Once, where it fits. There is no density target and there has not been for years.
12. Structure with real headings. H2 for sections, H3 for subsections, in hierarchy. Headings that exist for keywords rather than for navigation are obvious to everyone.
13. Keep paragraphs to two to four sentences. Most readers are on a phone, and a six-sentence paragraph is a wall there.
14. Make every claim specific. A number, a name, a version, a date. Specific claims are what gets quoted; general ones get skipped by readers and by assistants.
15. Cover the subtopics from check 7. Each one properly, in its own section if it deserves one.
16. Add internal links to related pages. Two to five, in context, with anchor text describing the destination. Not a list at the bottom. Our guide to AI SEO writing covers how the linking fits into the drafting process.
17. Link out where it genuinely helps. A source for a statistic, a tool being discussed. Sparingly and never to a direct competitor.
18. Include something visual. A screenshot, a chart, a table. Pages of unbroken text lose readers before the second scroll, and a table is often the fastest way to answer a comparison question. Our piece on how to embed YouTube videos without hurting page speed is worth reading before adding video.
Before you publish: 6 checks
19. Write the title tag for the click, not the algorithm. Under about 60 characters, containing the term, and giving a reason to choose you over the result above. Titles that read as keyword strings get skipped even when they rank.
20. Write the meta description as an advertisement. It does not affect ranking and it heavily affects clicks. Around 150 characters, stating what the reader gets.
21. Add alt text to every image. Describing the image, not stuffed with the term. This is accessibility first and SEO second, in that order.
22. Check it on a phone. Actually on a phone. Headings, tables and images all behave differently there and tables are where it usually breaks.
23. Check the page loads quickly. Large uncompressed images are the usual cause and the easiest fix.
24. Request indexing after publishing. Submit the URL in Search Console. It costs thirty seconds and it removes the wait for a crawl that might be weeks away.
The 2026 addition: writing to be cited
A growing share of searches now get answered by an assistant that read a handful of pages and summarised them. Being one of those pages is a different objective from ranking, and it is not difficult once you know what gets quoted.
Make statements that stand alone. A sentence that requires the previous three to make sense cannot be lifted into an answer. One that contains its own subject and a specific fact can.
Answer the question in the heading immediately below it. The pattern of question heading followed by direct answer is the most quotable structure there is.
Put facts in tables and lists. Structured information is extracted more reliably than the same information in prose.
Be specific and be checkable. Assistants favour pages that make verifiable claims, and vague marketing language is filtered out.
Include the boring detail. Prices, limits, versions, requirements. The specifics that a persuasive page leaves out are exactly what someone asks an assistant for.
What to stop doing
These were once advice, are still repeated, and now range from useless to harmful.
Keyword density targets. Never a real ranking factor and now a reliable way to make copy read badly.
Word count minimums. "Write 2,000 words" produces padding. Length should be whatever the question needs, and a 700-word page that answers completely beats a 2,500-word page that circles.
Exact-match repetition. Repeating the phrase in every heading is visible to readers and adds nothing.
LSI keyword lists. Cover the subtopics properly and the vocabulary follows. Sprinkling related words does nothing on its own.
Stuffing the first paragraph. Once, naturally, is the whole rule.
Publishing thin pages at volume. The cost of producing pages has collapsed, which has made publishing the wrong page the main risk rather than a minor one.
The checklist as a table
| Stage | Checks | The one that matters most |
|---|---|---|
| Before writing | 1 to 9 | Confirming intent from the actual SERP |
| While writing | 10 to 18 | Answering in the first hundred words |
| Before publishing | 19 to 24 | A title written for the click |
| Ongoing | Review at 90 days | Updating rather than republishing |
If you only adopt three of the twenty four, take those. They account for most of the difference between pages that rank and pages that do not.
After publishing
Check the page in Search Console after four to six weeks. Two numbers matter: whether it is getting impressions at all, and whether the clicks match the position.
Impressions but no clicks means the title and description are the problem, and rewriting those is a ten minute job with an immediate effect. No impressions at all means the term or the intent was wrong, which sends you back to checks 1 and 2.
Then review at ninety days and update rather than rewrite. Updating keeps the URL, the links and the history, which is the whole value of the page.
Where this gets automated
Most of the twenty four are mechanical and some are judgement. The mechanical ones automate well: structure, internal linking, metadata, indexing requests.
Distribb runs those. It researches the term, checks what ranks, drafts to the structure above, links the page into the rest of your site and publishes it. What it cannot do is check 5, the differentiator, because that has to come from something only you know, and a page without one is competent rather than distinctive. Our roundup of SEO article writer software covers the alternatives, and the best AI content creation tools compares the drafting side.
Frequently asked questions
How many times should the keyword appear? There is no number. Use it where it reads naturally, usually the title, one heading and the first paragraph, and stop thinking about it after that.
How long should an SEO article be? As long as the question needs. Look at what the top results cover and match the coverage rather than the word count. If the writing itself is the constraint rather than the checklist, our comparison of SEO article writing services covers what outsourcing costs.
Do meta descriptions affect rankings? No, and they affect clicks a great deal, which is why they are worth writing properly.
Does AI-written copy rank? Yes, when it is genuinely useful. What fails is thin content published at volume, which happens to be mostly generated, so the two get conflated.
How soon should I expect results? Weeks for long tail terms on an established site, months for competitive ones on a new domain. Check at four to six weeks for impressions rather than for rankings.
Use it as a gate, not a reference
Print the three groups and run a page through them before publishing rather than reading them beforehand. The checks that get skipped are always the first nine, and those are the ones that decide whether the rest of the work was worth doing.
If you want the mechanical two thirds handled on a schedule, Distribb's AI content writer runs the research, structure, internal linking and publishing, and leaves you the part that needs your own knowledge.