seo-keyword-to-page

Keyword to page: table-driven SEO planning

This skill runs keyword analysis as a fixed sequence of tables. Each table forces one decision into the open, so gaps show up as empty cells instead of hiding inside paragraphs. Work through the tables in order, fill every cell (or mark it n/a and log why in T9), and only then write the short commentary that connects them.

The core concept

SEO planning is five bets, answered in order. The numbers decide where to compete; the page decides whether you win.

Rows: the five stages of the plan. Columns: the question each stage answers, the tables that answer it, and what Google actually evaluates behind the tool metrics.

Stage	Question	Tables	What Google actually evaluates
Demand	Is anyone searching for this?	T1, T3	Nothing directly. Volume is a tool estimate; Google matches pages to queries by meaning. Ground truth is Search Console impressions.
Intent	What does the searcher want?	T2	The live results page is Google's current answer. What ranks is the brief.
Winnability	Can we beat page 1?	T4	Pages and links, among many signals. Google does not use DR, DA or KD; they are third-party proxies.
Page	What single page serves it best?	T5, T6	Helpful, people-first, non-commodity content with information the current results lack.
Packaging	Will they click?	T7, T8	Title links and snippets built by Google from your title, H1 and page text, often rewritten.

House rules

Rows: the output rules. Columns: the rule, and why it exists.

Rule	Why
Think in tables. Produce each stage's table before reasoning about the next stage.	Structure makes every decision explicit and comparable across keywords.
Caption every table with one line: Rows: ... Columns: ...	A reader must know what a single cell means without guessing.
Prose is connective only: at most two short lines between tables, plus a closing summary of three lines or fewer.	The analysis lives in the tables; prose only links them.
Never use em dashes. Use commas, colons or parentheses.	House style.
Never invent a metric. Every number carries a source tag. If data is unavailable, write n/a and still complete the qualitative columns.	A plausible fake volume is worse than a blank: it drives wrong priorities.
Describe tool metrics as estimates. Search Console is the only first-party performance data.	Google states that third-party tools have no access to its ranking data.
Read intent and page type from the live SERP, never from the keyword's wording alone.	The same words can hide opposite intents.
On how Search works, Google's documentation outranks third-party advice. Use practitioner studies only for magnitudes (click loss, rewrite rates).	Keeps the plan grounded in what Google says it does.
One country and one language per analysis, stated in T0.	Volume, difficulty and SERPs differ by market.

Source tags (put in a Source column, or in brackets after a value):

Rows: the tags. Columns: the tag and what it means.

Tag	Meaning
[Ahrefs] [Semrush] [KWP]	Third-party tool estimate (KWP = Google Keyword Planner)
[GSC]	Google Search Console, first-party data
[SERP]	Observed on the live results page
[User]	Provided by the user
[Judgment]	Reasoned assessment; allowed only for qualitative scores (business value, angle, information gain)
[Illustrative]	Made-up teaching data; never allowed in a real plan
n/a	Not available; the reason goes in T9

Step 1: pick the mode

Rows: request types. Columns: the tables required for that request.

Request	Tables required
Full keyword plan for a site or topic	T0, T1, T2, T3, T4, T5, T6, T7, T9 (T8 for priority pages, or on request)
"Is keyword X worth targeting?"	T0 (short), T2, T3, T4, T5, T9
Content brief for one page	T0, T1 (that cluster only), T2, T6, T7, T8, T9
Optimize an existing page	T0, T2, T7, T8 (including the cannibalization check), T9
Titles and meta descriptions only	T0 (short), T2, T7

Step 2: get the data

Rows: the inputs the tables need. Columns: preferred source, fallback, and what to do when neither is available.

Input	Preferred	Fallback	If unavailable
Search volume, traffic potential	Ahrefs Keywords Explorer (use a connector if one is available)	Semrush; Keyword Planner ranges	n/a; rank demand relatively from SERP and GSC evidence
Keyword difficulty	Ahrefs KD	Semrush KD% (never mix tools in one table)	n/a; judge from the T4 SERP columns
SERP composition (who ranks, page types, DR, linking domains)	Ahrefs SERP overview	Live web search of the query: read titles, URLs, domains	Never skip: at minimum read the top 10 titles and domains
AI Overview presence	Live SERP	Tool SERP-features column	unknown; treat informational queries as likely affected
Own rankings and clicks	Search Console	Ahrefs GSC reports	Ask the user for an export
Site strength	Ahrefs DR of own domain	Moz DA	n/a; compare against the SERP qualitatively

If a data connector is available (for example an Ahrefs connector with keyword overview and SERP overview tools), use it before asking the user. If no tool data is available, say so once in a single line, then deliver every qualitative column in full. If a code tool is available, use it for character counts and priority maths.

Step 3: fill the tables

T0 Brief

Rows: the brief fields. Columns: the value, its source tag, and its status.

Field	Value	Source	Status (Confirmed / Assumed / Missing)
Site or domain
Site strength (DR)
Market: country and language
Offer: what we sell
Ideal customer
Conversion action
Existing relevant pages
Seed topics or keywords
Data access (tools, GSC)
Constraints (budget, cadence, brand rules)

Ask only for fields that block the work (market and offer). Assume the rest, mark them Assumed, and carry them into T9.

T1 Keyword universe and clusters

Rows: individual keywords. Columns: the cluster each keyword belongs to and the evidence for the grouping.

Keyword	Cluster head	Shared top-10 URLs with head	Parent topic (Ahrefs)	Volume	Source

Rules:

• Expand seeds from tool suggestions, autocomplete, People Also Ask, Search Console queries and the customer's own words.
• Cluster by SERP overlap, not by similar wording. House threshold: 4 or more shared URLs in the top 10 means one page; 1 to 3 means decide by intent in T2; 0 means separate pages.
• One page per cluster. Never create a page per keyword variation: Google understands synonyms, and mass-producing variant pages to manipulate rankings breaches its scaled content abuse policy.

T2 Intent read (the three Cs)

Rows: clusters, one per cluster head. Columns: the intent and the three Cs read from the live top 10, plus the evidence.

Cluster head	Dominant intent	Content type	Content format	Content angle	SERP features	Evidence (top 3 results)

Rules:

• Intent: informational, commercial investigation, transactional or navigational.
• Content type: blog post, product page, category page, landing page, tool, video.
• Content format: how-to, list, comparison, review, template, definition.
• Content angle: the selling point the top results share (free, for beginners, step by step, for small teams).
• Match type and format; differentiate on angle.
• Mixed SERP: record the split (for example "6 guides, 4 product pages"), then follow the majority or send the cluster to Skip.
• SERP features to record: AI Overview, People Also Ask, video, local pack, shopping, forums.

T3 Demand

Rows: clusters. Columns: raw demand, AI-adjusted demand and business value.

Cluster head	Head volume	Traffic potential	Trend or seasonality	AI Overview	Adjusted demand	Business value (1-3)	Source

Rules:

• Prefer traffic potential (organic traffic of the current #1 page across all its keywords) over single-keyword volume.
• Adjusted demand = traffic potential × 0.42 when an AI Overview appears on an informational query; otherwise adjusted demand = traffic potential. Basis: Ahrefs' December 2025 data links an AI Overview to 58% lower click-through at position 1. Re-verify this figure yearly.
• Business value comes before volume. A 90-search keyword with buying intent can beat 9,000 informational searches.

Rows: business value scores. Columns: the score and its meaning.

Score	Meaning
3	The searcher could buy what we sell now (product, pricing, comparison, alternatives)
2	Adjacent problem our product solves, one step from a purchase
1	Educational only, weak path to conversion

T4 Winnability

Rows: clusters. Columns: difficulty, what page 1 is made of, and our position against it.

Cluster head	KD (tool)	Linking domains implied	Top-10 DR range	Top-10 results from sites with DR ≤ ours	Linking domains to ranking pages (min to median)	Our topical depth (existing pages on the topic)	Winnability (1-3)

Rows: the linking domains Ahrefs estimates a page needs for the top 10. Columns: Ahrefs KD values. Use only with Ahrefs KD.

KD	0	10	20	30	40	50	60	70	80	90
Linking domains needed	0	10	22	36	56	84	129	202	353	756

Rows: winnability scores. Columns: the score and the condition for it.

Score	Condition
3	Several top-10 results come from sites no stronger than ours, and we can match the ranking pages' linking domains soon
2	The gap is closable within about 6 months with supporting content and link building
1	Page 1 is dominated by far stronger sites whose pages carry hundreds of linking domains

KD estimates the top 10 (not the top 3), ignores on-page quality and is identical for every site. It never decides winnability on its own; the SERP columns do.

T5 Verdict and priority

Rows: clusters, sorted by priority, highest first. Columns: the three scores, the priority, the verdict and the next action.

Cluster head	Business value	Winnability	Demand score	Priority (BV × W × D)	Verdict	Reason (15 words max)	Next action

Demand score: 3 = top third of adjusted demand within this set, 2 = middle third, 1 = bottom third.

Rows: verdicts. Columns: the scoring condition and the action.

Verdict	Condition	Action
Go now	Business value ≥ 2, winnability 3, demand score ≥ 2	Publish or optimize now
Quick win	Business value ≥ 2, winnability 3, demand score 1	Publish fast and link it to a money page
Build toward	Business value 3 with winnability 1 or 2, or business value 2 with winnability 2	Build the supporting cluster and links first; recheck in 3 to 6 months
Skip	Business value 1, or business value 2 with winnability 1, or an intent we cannot serve	Leave out and note it in T9 (keep only if the plan explicitly includes awareness content)

T6 Page map

Rows: pages, one per cluster with a Go now, Quick win or Build toward verdict. Columns: where the page lives and what it must do.

URL	Page type	New or existing	Primary keyword	Secondary keywords	Main concept (one sentence)	Information gain	Links to	Linked from

Rows: intents from T2. Columns: the page type that serves each intent and where it lives.

Intent	Page type	Lives in
Transactional	Product, pricing, sign-up, category page	Main navigation
Commercial investigation	Comparison, alternatives, use case, industry page	Main navigation (compare or solutions section)
Informational	Guide, how-to, template, glossary	Blog or resources directory
Navigational (own brand)	Home, product, login, support	Main navigation

Rules:

• Main concept sentence: who the page is for, the question it answers, and what it adds beyond the current top 10.
• Information gain must name something concrete: original data, first-hand testing, real screenshots, worked examples, a tool or a template. "Better written" does not count.
• Existing page first: if a page already targets the intent, optimize it instead of creating a new one.
• Cannibalization: merge only pages that serve the same intent. Pages serving different intents can both rank.
• Every blog post links to its navigation page with descriptive anchor text; every navigation page links to its supporting posts.

T7 SERP packaging

Rows: pages from T6. Columns: title, H1, meta description and slug, with exact character counts and checks.

Page	Title tag	Title chars	H1	Meta description	Meta chars	Slug	Checks

Rows: the packaging elements. Columns: the rule and the target.

Element	Rule	Target
Title	Primary keyword early, one clear promise, no repeated words, no sitewide boilerplate, brand once at the end on navigation pages	About 50 to 60 characters. Google sets no limit but truncates to device width (about 600px on desktop).
H1	The most prominent text on the page, carrying the same promise as the title	One visible main heading
Meta description	Unique and specific: outcome, key facts (price, author, specs), soft call to action	About 150 to 160 characters. Google sets no limit.
Slug	Short readable words, lowercase, hyphens, no dates on evergreen pages	For example /freelancers or /blog/late-payment-reminder-email
Year in title	Only if the content is genuinely re-checked for that year	Google treats cosmetic date changes as a warning sign

Count characters exactly, with code if available; never estimate. Google rewrites most meta descriptions and many titles, so the page's opening paragraph must carry the same promise as the tags.

T8 On-page optimization checklist

Rows: individual checks, grouped by area. Columns: the area, the check, its status, the fix, and the basis (Google = Google documentation, Research = published practitioner data, Practice = common expert practice without official backing).

Area	Check	Status (Pass / Fix / n/a)	Fix	Basis
Packaging	Title unique, keyword early, within about 600px			Google; Practice for length
Packaging	One dominant H1 matching the title			Google
Packaging	Unique, specific meta description			Google
Packaging	Descriptive slug; visible breadcrumbs with BreadcrumbList markup			Google
Content	Direct answer or offer in the first screen			Google (snippets come from page text)
Content	Concrete information gain versus the current top 10			Google
Content	Clear sections, descriptive headings, tables for comparisons			Google
Content	Length set by the question, not by a word count			Google
Content	Byline linked to an author page; sources cited			Google
Content	Visible dates; date changed only with a substantial update			Google
Content	Relevant images placed near their text, with descriptive alt text			Google
Links	Hub and spoke links between the navigation page and its posts			Practice
Links	Descriptive anchor text in crawlable <a href> links			Google
Links	Key pages within three clicks of home; no orphan pages			Practice
Links	One URL per piece of content (301 first, rel="canonical" second)			Google
Links	Cannibalization: Search Console, filter the query, Pages tab; two URLs splitting one intent means merge			Practice
Technical	Returns 200, not blocked by robots.txt or the CDN, no stray noindex			Google
Technical	Core Web Vitals at the 75th percentile: LCP ≤ 2.5 s, INP ≤ 200 ms, CLS ≤ 0.1			Google
Technical	Same main content on mobile and desktop			Google
Technical	Key content present as text in the rendered HTML			Google
Technical	No interstitial or ad blocking the main content			Google
Structured data	Organization, BreadcrumbList, Article (posts), SoftwareApplication or Product (product pages); markup matches visible text			Google
Structured data	No FAQ or HowTo markup added for search appearance (FAQ rich results retired 7 May 2026; HowTo in 2023)			Google, via trade press
AI search	Page indexed and snippet-eligible; site included in generative AI features in Search Console			Google
AI search	No effort spent on llms.txt, content chunking or AI-specific rewrites for Google			Google
Measurement	Generative AI performance report and Performance report reviewed			Google
Measurement	Changes judged only after recrawl plus a few weeks			Google

T9 Assumptions, gaps and provenance

Rows: every assumption, data gap and judgment call made in the tables above. Columns: what it is, what it affects, how much, and how to close it.

Item	Type (Assumption / Data gap / Judgment)	Affects	Impact (High / Medium / Low)	How to close

Step 4: deliver

Order: one line stating scope, market and data sources; then the tables in T-number order; then a closing summary of three lines or fewer naming the top three pages to build and why. If the user wants a document or web page, render the same tables there with the same captions; the tables stay the backbone.

Step 5: self-check before sending

Rows: the checks. Columns: the check and whether it passes.

Check	Pass?
Every table has a Rows/Columns caption
Every metric has a source tag or n/a
No invented numbers; [Illustrative] data appears only in teaching examples
Intent and page type come from the live SERP
One page per intent; no variant pages
Every cluster has a verdict
Title and meta character counts are exact
Tool metrics described as estimates, not Google facts
All assumptions and gaps listed in T9
No em dashes anywhere

What not to do (Google's documented position)

Rows: common SEO beliefs. Columns: the belief and what Google's documentation says.

Belief	Google's position
A higher DR or DA ranks you higher	Google says it does not use Domain Authority, and third-party tools have no access to its ranking data
There is an ideal word count	No minimum or maximum; length alone does not matter
Keywords in the URL or domain boost rankings	Hardly any effect beyond breadcrumbs
Headings must follow strict order and count	Order and count do not matter to Google Search (semantic order helps screen readers)
Meta keywords matter	Google does not use the keywords meta tag
Duplicate content triggers a penalty	Not a spam violation; consolidate for users and crawl efficiency
E-E-A-T is a ranking factor	Not by itself; ranking systems use factors that align with it
AI Overviews need llms.txt, chunking or special schema	None needed; the same SEO fundamentals apply
Publish a page for every keyword variant	Unnecessary, and at scale it breaches the scaled content abuse policy

Example of the expected format

T5 for a hypothetical invoicing SaaS with DR 35. Every value is [Illustrative].

Rows: clusters, sorted by priority. Columns: the three scores, the priority, the verdict and the next action.

Cluster head	Business value	Winnability	Demand score	Priority	Verdict	Reason	Next action
[competitor] alternatives	3	3	2	18	Go now	Buying intent; DR 40 to 60 blogs rank with few links	Build the /compare page
invoicing software for freelancers	3	2	3	18	Build toward	Money term; KD 42 implies roughly 60 linking domains	Publish 4 supporting posts first
contractor invoice template	2	3	2	12	Go now	Template intent, weak SERP, natural product tie-in	Publish the template post
late payment reminder email	2	3	1	6	Quick win	Forums rank; links to the reminders feature	Publish and link to the feature page
invoice	1	1	3	3	Skip	Mixed intent, dominated by the largest brands	None

Evidence base (verified September 2026; re-verify yearly)

Rows: the facts this skill relies on. Columns: the fact and its source.

Fact	Source
No title length limit; truncated to device width; title links built from the title, H1 and prominent text	developers.google.com/search/docs/appearance/title-link
No meta description limit; snippets come mainly from page content	developers.google.com/search/docs/appearance/snippet
Keyword variations, URLs, duplicates, headings, word count and E-E-A-T myths	developers.google.com/search/docs/fundamentals/seo-starter-guide
Information gain tests, date changes, bylines, E-E-A-T	developers.google.com/search/docs/fundamentals/creating-helpful-content
AI search runs on core ranking; no llms.txt or chunking; scaled content abuse; Search Console inclusion and reporting	developers.google.com/search/docs/fundamentals/ai-optimization-guide
Third-party tools have no access to Google's ranking data	developers.google.com/search/docs/fundamentals/third-party-seo
KD method and the KD-to-linking-domains table	ahrefs.com/keyword-difficulty
Three Cs of search intent	ahrefs.com/seo/glossary/search-intent
Traffic potential definition; limits of search volume accuracy	help.ahrefs.com (Traffic Potential; search volume accuracy)
AI Overview linked to 58% lower position-1 click-through (December 2025 data)	ahrefs.com/blog/ai-overviews-reduce-clicks-update
Google rewrote meta descriptions 62.78% of the time	ahrefs.com/blog/meta-description-study
FAQ rich results retired on 7 May 2026	Google FAQPage documentation notice, reported by trade press