Day 31. The blockers are fixed, the baseline is written down, and the keyword map says which pages exist and which are missing. For the next thirty days the job is to publish the missing ones in an order that builds on itself: one subject at a time, on a cadence you can keep, with every page answering one search. The engine matters more than any single page: a routine that survives to day 60 keeps producing after the plan ends.
Why one cluster at a time beats a spread of posts
Google scores relevance page by page, but it reads trust for a subject across your whole site. Internal links between related pages tell it which pages belong together and which is the hub, and Topic clusters explains why a site that covers a subject thoroughly tends to be treated as its authority. AI assistants behave similarly: How AI assistants choose what to cite shows they lean on sources that cover a question fully.
Ten posts on ten subjects give you ten orphans, each fighting alone. Five pages on one subject, linked to each other and to the page that makes money, give you a cluster that gets stronger with every addition.
The cadence is for you, not for Google. Nothing in Google's documentation rewards publishing frequency for its own sake; a steady rhythm turns the work into a habit and gives you a flow of data to learn from. Pick a rate you can hold for sixty days, such as two pages a week.
Week 5: brief the first cluster
Start with the cluster closest to money. The keyword map from Days 1 to 30: foundations already groups keywords by subject, and The buying-intent ladder tells you which group sits nearest a purchase.
Each cluster has a pillar page that catches the broad search (often a service or category page you already have) and supporting pages that answer the narrower questions around it. Search intent decides the shape of each supporting page: a question wants an answer, a comparison wants a comparison, a "how to" wants steps.
Write briefs, not drafts. A brief fits on one page: the primary keyword, the intent, the question the first screen must answer, the headings, the internal links in and out, the publish date and the writer. A writer with a brief produces a page that fits the cluster; a writer without one produces a blog post.
In place by day 37: briefs for the pillar and its supporting pages, each with one keyword and one intent, scheduled on the calendar.
Weeks 6 and 7: publish, link, repeat
Publish the pillar first, then the supporting pages in calendar order. For every page:
- Write to the brief, with the answer in the first screen, following Writing a page that ranks.
- Set the URL, title and meta description before it goes live, using URLs, titles and meta descriptions.
- Add the trust signals from E-E-A-T and the trust signals Google looks for: a named author, a date, sources for claims, and first-hand detail only someone who does this work would know.
- Structure the answer so it can be quoted, as Writing for AI citations describes: a direct answer up top, self-contained sections, plain claims.
- Link the pillar to the new page and the new page back to the pillar, and add a link from at least one older page Google crawls often, per Site architecture and internal links.
- Confirm the page is in the sitemap, request indexing through URL Inspection in Search Console, and add its keyword to rank tracking the same day.
Step five is the one most often skipped, and it's what makes a cluster a cluster rather than a folder. During week 7, while the first cluster finishes, brief the second so the engine never idles.
In place by day 51: the first cluster live and interlinked, every page tracked, and the second cluster briefed.
Week 8: check what Google did with the first cluster
The week 6 pages have had time to be crawled, so this is the first honest look. Open the Pages report in Search Console, as Google Search Console essentials walks through, and find each new URL.
A page marked "Crawled, currently not indexed" was fetched and set aside, which usually means it read as thin or as a near-copy of something already in the index. Go back to How Google ranks a page, work out which, and fix the page rather than waiting. A page that's indexed but appears in the Performance report for a different query than you briefed has an intent problem: the searcher wanted something else.
What you don't do this week is judge rankings: a page a fortnight old has no trend yet, and Reading rank movements honestly explains why the daily number is noise. Publish the second cluster on schedule, run the weekly review, and write down how many pages you shipped in four weeks. That count is proof the engine works.
In place by day 60: two clusters live, or one finished and one under way, every page indexed or on a fix list, tracking on every keyword, and a cadence held for four weeks.
What to take away
- Build one cluster at a time: a pillar, its supporting pages and the links between them, aimed at the subject nearest a purchase.
- A brief with one keyword, one intent and the links in and out is what turns a writer's draft into a page that fits the site.
- Internal links from the pillar, back to it and from an older page are what make a cluster work, and the step most often skipped.
- By day 60 the deliverable is a cadence you've held for four weeks, with every page indexed or on a fix list.
Next
Days 61 to 90: authority and AI proof earns the links and mentions that make other sites and AI assistants point at what you've built.