Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

SEO › Keyword Research

Keyword Mapping to URLs

Step-by-step keyword mapping: match clusters to ranking pages, decide keep, improve, consolidate or create, assign one URL each, and maintain it.
Tutorialintermediate7 min readUpdated

Before you start

  • The clustered list from the clustering guide, with intent labels, head terms, variants, demand estimates and the results page's dominant format.
  • Search Console (GSC) export of queries and pages with impressions, clicks and position over a long enough window to cover seasonality.
  • A crawl of the site, or at least a list of every indexable URL with its title, H1 and page type.
  • The site's priorities: which products, services or topics matter most this year, from the business.
  • A spreadsheet or database that will hold the map. Columns at minimum: cluster id, head term, intent, demand, format, current best URL, current position, decision, owner URL, page type, proposed title, priority, status, notes, last reviewed. The keyword map template sets these columns out with an instruction for each and a filled section to copy.
  1. Match each cluster to what currently ranks

    For every cluster, find the site's best-performing URL from the Search Console data: the page with the most impressions or the best position for the cluster's head term and variants. Record it with its position. Many clusters will have no ranking URL; some will have several, which is the signature of cannibalisation and gets a note. Join the crawl data so the current URL's title and page type sit next to the cluster's intent and format.

  2. Classify the current situation

    Each cluster now falls into one of a few states: no page; one page that fits (right intent and format); one page that does not fit (a product page for a how-to query, or a thin page); several pages competing; or a page that covers this cluster and others. Label the state, because it determines the decision in the next step.

  3. Decide: keep, improve, consolidate, split or create

    Apply a decision per state:

    StateDecision
    No pageCreate; record a proposed URL, title and page type from the cluster's format
    One page, fits, ranks wellKeep; the page owns the cluster
    One page, fits, ranks poorlyImprove; the page owns the cluster and goes into the brief queue
    One page, wrong type or intentCreate the right page; decide whether the existing page keeps another cluster or is retired
    Several pages competingConsolidate into one owner (redirect the rest) or differentiate them onto distinct clusters
    One page covering several clustersKeep if the clusters are close siblings and the page serves them all; split if it is stretched

    The reader test from the cannibalisation guide applies: would one page serve the searcher better than a choice between two?

  4. Assign exactly one owner URL per cluster

    Write the owner URL into the row — existing, or proposed for new and rebuilt pages. Two rules are absolute: one cluster, one URL; and one URL may own several clusters only when they share intent and the page genuinely covers them. Check the proposed URLs against the site architecture so that new pages sit in the right section with the right path, and against the list of existing titles so that no two pages carry the same title or target.

  5. Record the page's intent reading and proposed title

    For each owner URL, note the dominant results-page format (SERP) from the cluster record and draft the title from the head term, in the pattern the title tag guide sets. The proposed title is what the brief and the page will use; writing it now surfaces duplicates and awkward clusters early.

  6. Prioritise

    Score each row on relevance to the business's current priorities, demand, the gap between current and achievable position, the strength of the competition on the results page, and the effort of the decision (a title change is cheap; a new guide is not). The score orders the brief queue. Transactional and commercial clusters close to the offer usually lead; informational clusters that build the topic's authority follow; the topic clusters guide explains why both are needed.

  7. Resolve consolidations and splits on the site

    Before new content, execute the structural decisions: redirect losing duplicates to the owner (never a canonical tag (Canonical Tag) where a redirect is possible), update internal links, and split stretched pages into their owners with redirects where URLs change. The cannibalisation guide has the execution sequence. The map's status column tracks each. When the stretched page is one live page doing two jobs, the tutorial on splitting a page that serves different intents works through which job keeps the address, what moves to the new page and how the pair is told apart.

  8. Brief from the map

    Each row marked create or improve becomes a content brief, which pulls the cluster, the intent and format, the proposed title and the owner URL straight from the map. The brief adds what the page must cover and link to; the map stays the source for what the page is for.

  9. Wire internal linking and measurement to the map

    Internal links use the map: anchors for a cluster's head term point at its owner URL and nowhere else — the internal linking guide covers anchor discipline. Measurement uses it too: the SEO KPIs guide tracks each priority cluster's owner URL for its head term, so that progress is read per row rather than from site totals.

  10. Maintain the map

    On every publish, merge or removal, update the row. At the quarterly review: fold in new queries from Search Console, re-cluster where results pages have merged or split, re-check that no cluster has grown a second ranking URL, re-prioritise, and date the row. The map is the site's planning memory; its history explains why each page exists. Queries that competitors rank for and the site does not reach the map through a competitor keyword gap analysis, run at the same review.

Illustrative example — not a real result

A section of a garden-tools retailer's map

The cluster "best cordless hedge trimmer" (commercial; results page a mix of buying guides and category pages) currently has two ranking URLs: a blog post from three years ago and a category page, swapping positions. Decision: the blog post is rebuilt as the owner — a current buying guide — and the category page owns the transactional cluster "cordless hedge trimmers" instead, with each linking to the other with anchors that say which is which. "Hedge trimmer blade length guide" (informational) has no page and is created. "How to sharpen hedge trimmer blades" is owned by a help-centre article that fits and ranks reasonably; marked keep, low priority. The proposed titles are checked against the site's existing titles, and one duplicate with a product-category title is caught and reworded before any brief is written.

What the map is used for afterwards

  • Preventing cannibalisation: every new page proposal is checked against the map first.
  • Briefing: the brief's purpose comes from the row.
  • Internal linking: anchor-to-URL rules come from the rows.
  • Measurement: priority rows are the tracked set.
  • Architecture changes and migrations: the map lists every URL that matters and what it is for, which the migrations guide relies on.

A new product name has no search history, so launch SEO planning before demand exists maps problem and category queries first and reserves rows for branded queries that will appear later.

When the business also runs a YouTube channel, the tutorial on planning YouTube videos and web pages for the same searches adds an owner video beside each owner URL where the results page shows people want to watch, so neither side builds for a search without the other knowing.

Verify it worked

  • Filter the map by cluster. Each cluster worth ranking for has exactly one owner URL, existing or proposed.
  • Check any URL that owns several clusters. The clusters share intent and the page genuinely covers them.
  • Compare the proposed titles with the site's existing titles. No two pages carry the same title or target.
  • Request a URL that lost a consolidation. It redirects to the owner, and internal links point to the owner.
  • Open the brief for a create or improve row. Its cluster, intent, format, title and owner URL match the map row.

If it goes wrong

  • Two URLs still rank for one cluster. The consolidation was probably not carried out. Redirect the losing duplicate to the owner and update internal links.
  • An owner URL fits the cluster but ranks poorly. The row was probably marked keep. Change the decision to improve and send the row to the brief queue.
  • A product page owns a how-to cluster. The page type does not match the results-page format. Create the right page and decide what the old page keeps.
  • One page is stretched across unrelated clusters. The clusters are not close siblings. Split it into owners, with redirects where URLs change.
  • Writers publish pages that are not on the map. The map is probably not reaching them. Tell writers and developers, and update the row on every publish, merge or removal.

Next steps

Common mistakes

  • A map with volumes and no decisions. A spreadsheet, not a plan.
  • Several owners for one cluster. The problem the map exists to prevent.
  • Mapping to pages that do not fit the format. A product page cannot own a how-to cluster.
  • Skipping the Search Console join. The site's existing rankings are the cheapest wins and the hidden conflicts.
  • Rebuilding annually. History and decisions lost.
  • Not telling the writers and the developers. The map only works if briefs, links and redirects follow it.

Frequently asked questions

What if a cluster has no suitable existing page?

The map records it as a new page with a proposed URL, title, page type and priority. The brief is written from that row. Until the page exists, the row's URL is planned, and the map flags it so that internal linking and measurement are set up when it goes live.

What if one page currently ranks for several clusters?

Decide whether it should. A guide legitimately covering a subject may own two or three closely related clusters; a page ranking weakly for five unrelated clusters is doing none of them justice and should be split. The map records the decision and the resulting URLs.

How often is the map updated?

Every time a page is added, removed or merged, and at the quarterly research review when new queries and cluster changes are folded in. A map that is rebuilt from scratch each year loses its history; one that is maintained becomes the site's planning record.

Terms used on this page

Frameworks

Where this page is used

Part of these learning paths

Sources

  1. Performance report (Search results) — Google Search Console Help

Related