Cluster Rationale · Marrowfield Payroll · cluster 7 of 214
Running payroll for the first time
Keywords in cluster
27
Sum of the rows
29,400 a month
Close variant groups
6
Actual demand
5,680 a month
Where the other 23,720 went
Group
Keywords
Each reports
Counts as
VG-01
7
2,400
2,400
VG-02
6
880
880
VG-03
3
1,000
1,000
VG-04
3
720
720
VG-05
2
480
480
VG-06
6
200
200
Why
Google defines average monthly searches as the figure for a keyword and its close variants. Seven phrasings of the same question all report 2,400 because they are all reporting the same searches. Adding them gives 16,800, and 14,400 of that is demand that does not exist.
The tell is on the sheet already: identical figures on rows that read differently are one group, not seven data points.
What it changes
Plan sized on the rows
1,120,000 a month
Plan sized on the groups
386,000 a month
A plan built by adding up rows is sized on the first number. It is 2.9 times the second one.
Page 1 of 1178 words100%
How to cluster keywords into a content plan
Keyword clustering means grouping the keywords from a research export into sets that one page can answer. Each set then gets a decision: a new page, a section of another page, an update to a page that already ranks, or no page at all. A good plan counts each search once, groups by what the searcher wants rather than by shared words, and lists pages in the order you'll build them.
The principle that matters most: plan pages, not phrases. One strong page can rank for hundreds of wordings of the same need, so the real work is deciding which searches belong together.
Check what your site already ranks for before you plan new pages. Photo: Myriam Jessier on Unsplash
Pull the export and read its columns
Seed the export with terms your customers use, plus your own site and a competitor's, and set the location and language you serve. Pull 12 months of monthly figures so seasonal peaks show up, and export Search Console data by query and by page for the same period.
Then read each column the way its maker defines it. Google's Keyword Planner help defines average monthly searches as searches for a keyword and its close variants1, averaged over 12 months by default. Its Competition column counts advertisers, so it describes the ad auction, not how hard it is to rank in unpaid results.
Four Keyword Planner columns, read correctly
Column
What Google says it measures
Use it for
Avg. monthly searches
Searches for the keyword and its close variants
Sizing, after merging variant rows
Competition
Advertisers that showed on it, relative to all keywords
Nothing in an organic plan
Top of page bid (high range)
Roughly the 80th percentile of what advertisers paid to show at the top
Judging what a search is worth
Organic impression share
How often a listing from your site showed
Spotting coverage you already have
Other keyword tools estimate volume their own way, so check their definitions before adding rows.
Cut what you won't build for
Much of a large export is demand you can't or shouldn't serve. Cut it before clustering, because every later step is faster on a shorter list. Log each cut with its reason on a separate sheet, so nobody adds the keyword back next quarter.
Your own brand terms, which existing pages already serve
Competitors' brand terms, which need their own plan and claims review
Keywords with no searches across the full 12 months
Products or services you don't sell
Places you don't serve
Searchers who will never buy, such as job seekers researching salaries
Difficulty scores and a big brand on page one aren't reasons to cut; they change when you build a page, not whether. If you serve one local area, keep the searches that name it, and check your Google listing and reviews with a local SEO audit.
Cluster keywords by intent, not wording
Two keywords belong in one cluster when one page could satisfy both searchers. Google's guidelines for its search quality raters describe four kinds of query intent2, including Know (find information) and Do (get something done, such as download or buy). They note that "barack obama height" has the same intent as "how tall is barack obama".
Matching on shared words fails in both directions. It merges "payroll journal entry" with "payroll journal entry template", though one searcher wants an explanation and the other wants a file. It splits "how to run payroll" from "running payroll step by step", which one page answers. The quickest check is the results page: when Google ranks the same pages for two keywords, they usually belong together.
Four working intents for a content plan, and the page each needs
Intent
The searcher wants
The page it needs
Definition
To know what something is
A clear answer near the top, then depth
How-to
To complete a task
Every step, in order
Artifact
A template, checklist or calculator
The thing itself first, explanation after
Commercial
To compare options or buy
Usually a product or comparison page, not an article
Size each cluster without double counting
Because each row's figure includes its close variants, rows that are variants of one another largely report the same searches. Google's Ads help says close variants3 include misspellings, singular or plural forms, reordered words and synonyms. Add those rows up and you count the same searches several times.
So merge before you add: group rows that are variants of each other, keep the highest figure in each group, then add the groups. Identical figures on rows that read alike are the giveaway. Keep both totals on the sheet so the gap stays visible.
ExampleSizing one clusterAn illustrative example written for this guide. The search volumes are invented.
how to run payroll: 2,400
how to run a payroll: 2,400
how to do payroll: 2,400
payroll process steps: 880
Summed: 8,080 searches a month. Merged: 2,400 plus 880, so 3,280.
The giveaway. Three phrasings carry the same 2,400, so treat them as one group of searches.
The number to plan on. The merged figure is about 40% of the sum. Size and order the plan on it.
Check the season next. A 12-month average hides peaks, so look at the monthly figures for anything tied to a date, such as year-end tasks or tax deadlines. Give those clusters a publish-by date well before the peak. Google's SEO Starter Guide says a change can take from a few hours to several months4 to show up in search results.
Check what you already rank for
A cluster with a page that already earns impressions isn't a gap. It's a decision about that page. In Search Console's Performance report, impressions count how often your site appeared in results5, and click-through rate is clicks divided by impressions. Filter the report to a cluster's main keywords, then open the Pages tab to see which URL earns them.
Keep figures from the Queries and Pages tabs apart. Search Console counts query data per site and page data per page, so their totals can differ.
Then check the average position of each page that earns a cluster's impressions. A page ranking in the top 10 with plenty of impressions but very few clicks often answers the wrong question, such as an article where searchers want a template. Changing the page type is usually cheaper than writing a new page. Decisions about the rest of your old pages belong in a content audit.
Turn clusters into pages, in build order
One page can rank for many searches. Ahrefs, which sells SEO software, studied 3 million searches in 2017. The average top-ranking page also ranked in the top 10 for nearly 1,000 other relevant keywords6, and the median page for about 400. So fold clusters that share an intent at different depths into one page, and split a cluster only when it mixes two intents.
Resist one page per phrasing. Google's spam policies describe pages created to rank for specific, similar search queries7 as doorway abuse. Give every cluster a page count, a page type and a verdict: keep, expand, new page, fold into another page, or drop.
1
Fix pages that rank but miss
Top-10 pages with few clicks often need a different page type, not more words.
2
Publish seasonal pages early
Order them by publish-by date, set months ahead of each peak.
3
Expand thin pages that rank
An expansion builds on the links and rankings the page already has.
4
Add new pages last
Order them by merged demand and top of page bid together.
Before you share the plan
Every cluster counts each group of variants once
Seasonal clusters have a publish-by date
No decision rests on the Competition column
Every cluster has a page count, a page type and a verdict
Clusters you already rank for name the URL
Every cut keyword is logged with its reason
Next, brief each page with an SEO content brief template before anyone writes, and fit the order to your team in an editorial calendar. To run the whole method on your own export, open the template at the top of this page in River.
How to use this template
Use the files on their own: the five documents explain the method, and the four sheets hold the keywords, the cuts, the page map and the gaps. They arrive filled in with a made-up payroll company's plan, so clear its rows before adding yours.
Open it with Edit with AI and send River your keyword export. Before clustering anything, River puts the merged demand beside the summed figure, flags seasonal terms and asks what you sell, where and to whom. Then it fills every sheet, checks coverage against any Search Console data you send and orders the pages. Where a column is missing, it says which decision is running without it.
1
Download it or open it in River
Download the nine files free, five Word documents and four CSV sheets, or open the whole set in River.
2
Read and trim the export
Cut what you won't build for and merge variant rows, following Clustering Method, or let River do it from your file.
3
Cluster and map the pages
Group keywords by intent in Keyword Clusters, then give each cluster a page count, type and verdict in Cluster to Page Map.
4
Build in band order
Work through Coverage Gap and Plan Sequencing, fixing pages that already rank before writing new ones.
What's in the template
Two hundred and fourteen clusters, a hundred and sixty-four pages, and a build orderThe Keyword Clusters sheet, the Cluster to Page Map and the Coverage Gap in build order.
Keyword Clustering Template and Page Plan
Keyword Clusters
Calibri11
Keyword Clusters
Marrowfield Payroll, a fictional payroll product. Export pulled 12 August 2026, 18,400 ideas cut to 6,120, joined to Search Console for 1 August 2025 to 31 July 2026.
Keyword
Cluster
Intent
Group
Avg
Peak
Bid
Impr. share
Google agrees
how to run payroll
Running payroll first time
Procedural
VG-01
2,400
steady
4.10
18%
Yes
running payroll step by step
Running payroll first time
Procedural
VG-01
2,400
steady
4.10
18%
No, wording
payroll process steps
Running payroll first time
Procedural
VG-02
880
steady
4.40
11%
Yes
payroll journal entry
Payroll journal entries
Definitional
VG-08
1,900
steady
6.20
0%
Yes
payroll journal entry template
Journal entry template
Artifact
VG-09
1,300
steady
7.80
4%
No, intent
payroll year end checklist
Payroll year end
Procedural
VG-14
1,300
9,400 Mar
5.60
22%
Yes
what is a p60
P60 explained
Definitional
VG-21
14,800
31,000 Apr
2.10
9%
Yes
payroll software
Payroll software category
Commercial
VG-30
40,500
steady
24.50
31%
Yes
best payroll software small business
Payroll software comparison
Commercial
VG-31
8,100
steady
31.20
2%
Yes
payroll reconciliation template
Reconciliation template
Artifact
VG-44
720
steady
9.10
6%
Yes
gross to net pay calculator
Gross to net calculator
Artifact
VG-51
12,100
steady
5.40
0%
Yes
p11d deadline
P11D filing
Procedural
VG-63
3,600
8,800 Jun
2.90
0%
Yes
The four intents, and what each one decides
Intent
Searcher wants
Page type it forces
Definitional
To know what a thing is
Explainer, or a reference table at scale
Procedural
To know how to do it
Ordered procedure. Completeness beats depth
Artifact
An object: template, checklist, calculator
The object above the fold, prose after
Commercial
To evaluate or buy
Usually not a content page at all
Google's own concept grouping disagreed with these clusters 31 times out of 214. The pattern: when the disagreement is about intent it was right, 19 times. When it was about wording, we were.
Cluster to Page Map
Calibri11
Cluster to Page Map
Every cluster gets a page count, a page type and one of five verdicts. 214 clusters resolve to 88 existing pages and 76 new ones.
Cluster
Sum of rows
Real
Owner
Owner rate
Type
Pages
Verdict
Band
Reconciliation template
4,180
1,360
/blog/payroll-reconciliation-guide
0.4%
Artifact
1
Expand
1
Journal entry template
3,100
1,300
/blog/payroll-journal-entries
0.6%
Artifact
1
Expand
1
Payroll year end
18,700
4,100
/blog/payroll-year-end
1.9%
Dated
1
Expand
2
Statutory sick pay
14,200
7,900
none
n/a
Procedure
1
New page
2
P60 explained
42,800
19,600
/glossary/p60
1.1%
Explainer
1
Expand
3
Running payroll first time
29,400
5,680
/guides/how-to-run-payroll
3.1%
Procedure
1
Expand
3
Gross to net calculator
21,600
12,100
none
n/a
Artifact
1
New page
4
Payroll journal entries
4,900
1,900
none
n/a
Explainer
1
New page
4
Payroll software comparison
14,800
8,690
none
n/a
Comparison
1
New page
4
Payroll tax rates
7,200
4,400
/resources/payroll-tax-rates
6.7%
Reference
1
Keep
n/a
Payroll for contractors
2,100
880
none
n/a
section
0
Fold
5
Payroll software category
58,900
40,500
/product
2.4%
Category
0
Drop
n/a
Where the 214 clusters went
Verdict
Clusters
Pages
Why
Keep
34
34 existing
Owner is adequate against the real demand
Expand
54
54 existing
Owner is thin or is the wrong type. Inherits what the page already earns
New page
67
76 new
No owner. Nine split because the cluster carries two intents
Fold
47
0
Same intent at a different depth. Ships as a section
Drop
12
0
Off-product, off-geography, or a query a product page should own
The 40,500 row is the largest in the export and it is a Drop. A content page targeting a category term competes with the page that converts, and if it wins, the company has replaced a page that sells with one that does not.
Coverage Gap
Calibri11
Coverage Gap
The 67 clusters with no owner, in build order. Sequenced on real demand and commercial evidence together, with seasonal clusters pulled to their ship-by date.
#
Cluster
Real demand
Peak
Bid
What page one serves
Owner
Ship by
1
Statutory sick pay
7,900
11,200 Jan
3.30
Two government pages, three long procedures
Content
2026-11-10
2
Gross to net calculator
12,100
steady
5.40
Five interactive calculators, no articles
Product
2026-11-24
3
Payroll journal entries
1,900
steady
6.20
Three explainers, one textbook extract
Content
2026-12-01
4
Payroll software comparison
8,690
steady
31.20
Four vendor comparisons, two aggregators
Sales
2026-12-08
5
Auto enrolment duties
5,400
steady
4.70
Two regulator pages, four long guides
Content
2027-01-19
6
Payslip requirements
3,200
steady
2.80
A table on one competitor, prose elsewhere
Content
2027-02-02
7
Holiday pay calculation
4,100
steady
3.60
Three calculators, two procedures
Content
2027-02-16
8
Payroll for directors
1,400
steady
8.90
One accountancy firm, four forum threads
Content
2027-03-02
Eight clusters came off this sheet before it was published
Where the coverage turned out to be
Clusters
A paginated archive nobody associated with the topic
3
A glossary page built in 2021
2
A support article outside the marketing site entirely
3
Search Console named an owner for 88 clusters. Organic impression share showed a listing from the site appearing for 96. All eight of the difference were about to be commissioned as new pages, and that column takes a minute to read.
Clustering MethodThe seven steps from a raw export to a page list, starting with what to cut.
Reading the ExportWhat each Keyword Planner and Search Console column measures, in Google's words, and which ones the plan uses.
Page Type GuidanceSeven page types, the evidence for choosing one, and the two type mistakes that cost the most.
Cluster RationaleThe worked example's contested calls, including the 31 clusters where Google's own grouping disagreed.
Plan SequencingThe five build bands, why fixes to pages that already rank come first, and a quarterly check.
Keyword Clusters sheetEvery kept keyword with its cluster, intent, variant group, peak month, bid and current coverage.
Discarded Keywords sheetEach discard rule with an example keyword, how many keywords it cut, why, and when to revisit.
Cluster to Page Map sheetOne row per cluster with summed and merged demand, owner URL, page type, verdict and band.
Coverage Gap sheetClusters with no page yet, in build order, with what ranks now and who builds each one.
Common questions
What is keyword clustering?
Keyword clustering is the practice of grouping keywords that a single page can answer, so you plan pages rather than chase phrases one at a time. Keywords belong together when one page would satisfy each searcher, which you can test by whether Google ranks the same pages for them. Shared words alone are a weak signal.
How do you calculate a keyword cluster's search volume?
Take the highest figure in each group of close variants and add those, not every row. Google's Keyword Planner counts a keyword's close variants in its average monthly searches, so a plural and a reworded phrasing can report the same searches. Summing all the rows counts those searches several times and overstates demand.
Should I target the keywords with the most searches?
Not by default. The largest rows in an export are often category terms that your product page should own. An article competing for them can pull searchers away from the page that sells. Rank clusters on merged demand, commercial value and whether you already have a page, not on the raw volume column.
How many pages should a keyword plan produce?
Usually fewer pages than it has clusters, and fewer new pages still. Clusters with the same intent at different depths fold into one page as sections, many land on pages you already have, and some are dropped. In the template's worked example, 214 clusters become 164 pages, and 88 of those already exist.
How often should you redo keyword research?
Pull a fresh export about every six months, and check the plan in Search Console each quarter. Demand shifts, so a plan sized on last year's average drifts out of date. Each quarter, see whether fixed pages gained clicks and whether new pages draw impressions but few clicks.
Is this keyword clustering template free?
Yes. You can download all nine files without signing up: five Word documents and four CSV sheets. They arrive filled in with a fictional payroll company's plan, so you can follow every step on realistic data before clearing it. Edit with AI opens the same files in a River workspace.
What does Edit with AI do?
It opens the template in a free River workspace and asks for your keyword export, plus Search Console data if you have it. Before clustering, River shows how far the summed volume overstates real demand and which terms peak in one season. It then fills the sheets and orders the pages, and it never makes up a volume, bid or ranking URL.
Chandler has spent the last six years working in AI, building tools for millions of users and Fortune 100 companies. In 2025, he co-founded River to build a new kind of operating system for AI agents.
Keyword Clustering Template and Page Plan