Topic Cluster Building for RRF

Fresh

This content reflects a synthesized workflow from Metehan Alp's RRF methodology as of 2025-2026.

A workflow for building topic clusters engineered to cross the RRF citation threshold. Instead of building clusters based on keyword volume alone, this approach uses RRF math to determine exactly how many sub-query rankings you need and builds the cluster to achieve them.

Workflow Overview

Step 1: Core Topic Selection

Choose the topic your cluster will target. This becomes the hub -- the central page that all supporting content points back to.

Selection Criteria

• Topic aligns with your core business offering
• Search volume justifies the investment (1000+ monthly searches combined across variations)
• You have genuine expertise (E-E-A-T must be real, not manufactured)
• Competitive gap exists (not dominated by a single authoritative source)

Output

• Core topic statement (1 sentence)
• Primary target query
• Business justification (how citations drive revenue)

Step 2: Sub-Query Mapping

Map 8-16 sub-query variations using the Query Fan-Out methodology.

Process

1. Start with your primary target query
2. Generate sub-queries across all 5 categories:

Category	Example for "project management for remote teams"
Related	"remote project management software"
Implicit	"team communication tools for distributed work"
Comparative	"Asana vs Monday for remote teams"
Procedural	"how to set up project management for remote team"
Contextual	"challenges of managing remote projects"

3. Validate by testing on Google AI Mode to see which sub-queries are actually generated
4. Aim for 8-16 validated sub-queries

Output

• Validated sub-query list (8-16 queries)
• Category labels for each
• Google AI Mode validation notes

Step 3: SERP Data Collection

Collect ranking data for each sub-query.

Process

1. For each sub-query, pull the top 60-100 results from Google
2. Use Screaming Frog, Semrush, Ahrefs, or the Query Fan-Out script
3. Record your domain's position for each sub-query (or "NR" if not ranking)
4. Record competitor positions for the same sub-queries
5. Note which competitors appear across multiple sub-queries

Output

Position matrix:

Sub-Query	Your Domain	Competitor A	Competitor B	Competitor C
Related 1	#23	#5	#12	NR
Implicit 1	NR	#8	#3	#15
Comparative 1	#45	#2	NR	#7
...	...	...	...	...

Step 4: RRF Score Computation

Calculate the fused score for your domain and each competitor.

Formula

S(d) = sum of (1 / (60 + rank)) for each sub-query where domain d ranks

Example Calculation

Suppose your domain ranks for 3 of 10 sub-queries: #23, #45, and #67.

S = 1/(60+23) + 1/(60+45) + 1/(60+67)
S = 1/83 + 1/105 + 1/127
S = 0.01205 + 0.00952 + 0.00787
S = 0.02944

This exceeds tau (0.020), so you meet the citation threshold.

Threshold Check

Refer to the practical rules:

Rule	Condition	Meets Threshold?
Rule 1	2 rankings in top 40	Check
Rule 2	3 rankings in top 90	Check
Rule 3	1 at #1 + 1 in top 80	Check
Rule 4	4 rankings in top 140	Check

Output

• RRF score for your domain
• RRF scores for top 3 competitors
• Gap to threshold (how far above/below 0.020)
• Specific sub-queries where improvement yields the highest score increase

Step 5: Content Gap Analysis

Identify which sub-queries you need to create or improve content for.

Process

1. List all sub-queries where you rank "NR" (not ranking) -- these are complete gaps
2. List sub-queries where you rank 60+ -- these are weak presence
3. List sub-queries where you rank 1-40 -- these are strengths to maintain
4. Calculate: if you moved each gap from NR to rank 30, what would your new RRF score be?
5. Prioritize gaps by score impact

Prioritization Framework

Priority	Condition	Action
P1 (Critical)	NR for sub-query, and fixing it would push total score above 0.020	Create dedicated content immediately
P2 (Important)	Rank 60-100, improvement to top 40 would add significant score	Optimize existing content
P3 (Maintain)	Rank 1-40	Monitor, do not deprioritize
P4 (Low impact)	Even rank #1 for this sub-query would not move the needle	Deprioritize

Output

• Prioritized gap list
• Score impact per gap (if fixed)
• Content creation vs. optimization decisions

Step 6: Cluster Architecture

Design the hub-and-spoke content cluster.

Hub Page

The hub is your main page targeting the primary query. It:

• Covers all sub-query topics at summary level
• Links to each spoke page for depth
• Is the page that should accumulate the highest RRF score
• Contains the strongest E-E-A-T signals

Spoke Pages

Each spoke targets 1-2 specific sub-queries. Spokes:

• Go deep on one sub-topic
• Link back to the hub
• Link to adjacent spokes where relevant
• Are self-contained but reference the hub's broader context

Architecture Map

Internal Linking Rules

• Every spoke links to the hub (mandatory)
• Hub links to every spoke (mandatory)
• Spokes link to 1-2 related spokes (when topically relevant)
• Anchor text uses the sub-query keyword naturally
• No orphan pages (every page has at least 2 internal links)

Output

• Cluster architecture diagram
• Hub page brief (title, target queries, sections)
• Spoke page briefs (one per gap from Step 5)
• Internal linking map

Step 7: On-Page Engineering

Optimize each page in the cluster for its target sub-queries.

Hub Page Optimization

• H1 matches primary query
• First 150 words contain the direct answer
• Each H2 previews a spoke topic and links to it
• Schema: Article + Organization + Person
• AI-directed HTML comments
• CC BY 4.0 licensing
• CiteMET share buttons

Spoke Page Optimization

• H1 matches the target sub-query
• First 150 words answer the sub-query directly
• ~500 tokens per major section
• Links back to hub in the introduction and conclusion
• Schema: Article + Person
• Answer-first structure for every section

Content Specifications

Spec	Hub Page	Spoke Page
Word count	2,500-4,000	1,200-2,000
H2 sections	6-10	3-5
Internal links	8-12	4-6
External citations	3-5	2-3
Schema types	3	2
Images	2-4	1-2

Validation

After publishing the full cluster, re-run Steps 3 and 4:

1. Wait 2-4 weeks for indexing and ranking changes
2. Re-collect SERP data for all sub-queries
3. Re-compute RRF scores
4. Test prompts on ChatGPT, Perplexity, and Google AI Mode
5. If score >= 0.020 and citations appear: monitor monthly
6. If score < 0.020: identify remaining gaps and expand the cluster

RRF Scoring Example

Full worked example for a 10 sub-query cluster:

Sub-Query	Your Rank	Score Contribution
"project management remote teams"	#15	1/75 = 0.01333
"remote PM software comparison"	#8	1/68 = 0.01471
"how to manage remote projects"	NR	0
"remote team collaboration tools"	#32	1/92 = 0.01087
"asana vs monday remote"	NR	0
"remote project challenges"	#55	1/115 = 0.00870
"remote team productivity tools"	#22	1/82 = 0.01220
"PM methodology for distributed teams"	NR	0
"remote onboarding project management"	#41	1/101 = 0.00990
"async project management best practices"	#18	1/78 = 0.01282

Total RRF Score: 0.08253

This far exceeds the threshold of 0.020. Even with 3 gaps (NR sub-queries), the breadth across 7 rankings produces a strong fused score.

If you only ranked #1 for the primary query:

S = 1/(60+1) = 0.01639

Below threshold. This demonstrates why breadth matters more than a single #1 ranking.

Further Reading

• RRF Top-n Playbook -- The mathematical foundation
• Query Fan-Out Analysis -- Sub-query identification methodology
• AEO Content Pipeline -- How to write each piece in the cluster