Recommended placement — where a keyword goes on your site

Summary#

Keyword tools have always answered is this keyword worth having — volume, difficulty, intent, cost per click. Recommended placement answers the question that follows it: where does this keyword go on my website. Every keyword result now names a page type, the existing page to use when one of yours already fits, and the on-page slots the keyword should occupy.

Purpose#

Knowing a keyword is valuable is only half a decision. The other half is which page should target it, and getting that half wrong wastes the work: a buying term written up as a blog post will not convert, and a how-to question dropped onto a category page will not rank. That judgement used to live in the customer's head, or in an agency's. It is now attached to every keyword the platform shows you.

Overview#

Placement appears in two shapes, from one engine.

In single-keyword results — Keyword Overview — a fuller block sits between the insights and the action plan, adding the reason for the recommendation, the list of on-page slots, and a caution when the numbers warrant one.

In Keyword Vault, a Where these keywords go card groups your whole saved vault into the pages it implies, largest combined search volume first. Several keywords pointing at one page are shown as one job, because that is what they are.

How it works#

The recommendation is calculated, not generated. It uses no AI call and costs no credits. A keyword table can run to hundreds of rows; a model call per row would be slow and would bill you for something a rule answers exactly. The same keyword with the same metrics always produces the same recommendation.

What decides the page type. Search intent is the primary signal, refined by the shape of the keyword itself:

What decides the URL. If you have a website selected in the switcher, the platform reads that site's sitemap and uses the real page list. A keyword whose meaning is fully carried by an existing page of the right kind recommends Optimize and names that page — a real URL you can open. Everything else recommends Create, which names the page type to build but no URL. It does not print a guessed path like /collections/... for a page that does not exist: one invented URL is the fastest way to lose trust in every real number beside it, so a URL is shown only when it can stand behind one, which today means a page you already have.

Two rules keep the destination honest. A page must carry every meaningful word of the keyword, so /collections/running-shoes is offered for running shoes but not for best running shoes or nike running shoes — those are different pages. And the page must be the kind of page the keyword wants, so a keyword that needs a comparison article is never pointed at a category page.

With no website selected, or a site with no reachable sitemap, you still get the page type, but no URL: with nothing to match against every keyword is a Create, and a Create names no path.

What decides the on-page slots. A new page can own its URL and title outright, so it is told to use the keyword in the URL slug, page title, H1, first paragraph and meta description. An existing page is being retrofitted, so it is told to use the title, H1, body copy and internal link anchors. It is never told to change a live URL.

The page index is cached for 7 days per domain, shared across the account. Site structure changes on the order of weeks, and the alternative is asking your own origin for its sitemap on every keyword report.