What each page should rank for

Every page on the site gets a podium of the terms it should own, scored by six named factors — and each recommendation reads back as a sentence you can argue with.

What it is

A model that marries pages to keywords. Each page has a role (home, product, pricing, guide, blog hub, post, utility) which decides what search intent belongs on it. Candidate terms are scored as fit × reach × demand^0.6 × season × proximity × exclusivity, each factor reported separately. Every page gets a verdict — on-target, no-target, cannibalised, off-brief, or not-seo — plus ordered actions.

Also called: keyword to page mapping · page targets · which page owns which keyword · keyword assignment

See it
What each page should rank for
Won
56%
Avg cycle
9d
Pipeline
$486k
JanSep
The podium for one page — winner plus two runners-up — each with its six factor values and the one-line 'why'. Rendered by src/components/platform/seo/PageInsightModal.tsx. Sample data — no customer information appears here.
How it works
  1. 1roleOf(path) classifies the page; ROLE_INTENT decides which search intents are allowed on that role.
  2. 2Terms explicitly pointed at the page (target_url) are treated as a standing decision and always appear.
  3. 3Terms the page actually ranks for today are added whether assigned or not, using the URL Google returned.
  4. 4Remaining universe terms are scored on topical word overlap, gated by intent, and the top four are recommended.
  5. 5FIT is a gate not a nudge; REACH discounts anything above the stage ceiling to 0.25; PROXIMITY rewards positions 4–20 at 1.35; EXCLUSIVITY drops a term already owned by another page to 0.4.
  6. 6The composite orders the list, an assigned term gets a small bump, and the top three become the podium.
Why we built it

The module states the gap: "The Website tab could already tell you what a page ranks for. It could not tell you what it OUGHT to rank for, which is the question that actually drives work: every page is either earning a term, targeting one it hasn't won yet, or drifting with no term at all — and until that is written down, 'improve the website' has no defined next action." The design is deliberately explainable rather than clever, "because a recommendation nobody can argue with is a recommendation nobody can correct."

The problem
  • Pages had no assigned target, so they could not be measured or improved.
  • Recommendations were opaque, so they could not be corrected.
  • Buying-intent terms were being suggested for pages that could never convert them.
Sound familiar?
What you get
Every page has a named term to own, with the reasoning attached
Six factors reported separately, so a wrong recommendation can be traced
Terms above the current stage are surfaced as locked, not hidden
What's inside

See it on your own jobs

Twenty minutes, your numbers, no slide deck. We’ll build one of your real buildings in front of you and send you the estimate link at the end — yours to keep either way.

or keep browsing features →