C
Content SOPKnowledge Base
Search
← All topics

Find Topics to Write: Research, PAA Mining, Clustering, and Gap Analysis

runnable

4-step topic discovery system: search trend research, PAA mining, topic clustering, and gap analysis. Outputs a prioritized topic list with article title suggestions, content type, and keyword target. Used to fill a content calendar from scratch.

topic-researchkeyword-researchpaacontent-planningtopic-clusters
Agent trigger phrases: what should I write about · find topics to write · content topic ideas · topic research · blog topic ideas · what topics to cover · content ideas for

Find Topics to Write: Research, PAA Mining, Clustering, and Gap Analysis

The correct sequence for topic discovery is: understand what people search, mine what they ask, cluster related questions, map against what already exists. This produces a prioritized list with zero guesswork.

Step 1: Seed Keyword Research

Start with 3-5 seed keywords that represent the core of the business or niche:

Sources for seed keywords:

  • Primary services or products offered
  • Geographic + service combinations ("roof repair [city]")
  • Problem-based searches ("how to fix [problem]")
  • Comparative searches ("[service] vs [alternative]")

For each seed keyword, expand to:

  • The primary keyword (high volume, competitive)
  • 3-5 long-tail variants (more specific, lower competition)
  • 2-3 question variants (PAA candidates)

Step 2: PAA Mining

Run PAA Researcher in Deep Mode for the top 2-3 seed keywords. (See paa-researcher-modes.mdx for full protocol)

From PAA mining, extract:

  • All questions at awareness stage
  • All questions at consideration stage (these have highest content ROI)
  • All questions at decision stage
  • Questions that competitors are NOT currently answering well

Quick identification of unanswered PAA:

  • Search the PAA question in Google
  • If position 1-3 results only partially answer the question, or the content is 3+ years old → gap confirmed
  • If position 1-3 results are from low-authority sites → opportunity confirmed

Step 3: Topic Clustering

Group related topics and questions into content clusters. Each cluster gets one primary article (the pillar) and 2-4 supporting pieces.

Cluster structure:

CLUSTER: [Topic name]
Primary keyword: [main keyword — pillar content]
Pillar article: [title + target word count]

Supporting pieces:
1. [Supporting keyword] → [content type: FAQ / comparison / how-to]
2. [Supporting keyword] → [content type]
3. [Supporting keyword] → [content type]

Internal link plan:
- Pillar links to all supporting pieces
- Supporting pieces all link back to pillar
- Supporting pieces link to each other where contextually relevant

Step 4: Gap Analysis Against Existing Content

Before adding any topic to the production queue, check:

  1. Does any existing page already target this keyword?
  2. Is the existing page ranking (positions 1-20)?
  3. Does the existing page fully cover the topic?

Decision matrix:

| Existing Content? | Ranking? | Action | |------------------|---------|--------| | No | N/A | Create new | | Yes | No | Update existing (refiner workflow) or rewrite | | Yes | Positions 4-20 | Update and expand existing page | | Yes | Positions 1-3 | Leave it — don't disturb a winner |

Step 5: Priority Output

Score each topic and output a production list:

TOPIC PRODUCTION LIST:

Tier 1 — Produce immediately:
1. [Title] | [Keyword] | [Word count] | [Content type] | Priority: [score]
2. ...

Tier 2 — Produce this quarter:
...

Tier 3 — Monitor / Low priority:
...

Tier criteria:

  • Tier 1: High commercial intent + easy to medium difficulty + no existing content
  • Tier 2: Informational with traffic potential + medium difficulty + cluster value
  • Tier 3: Long-tail informational + low volume + nice-to-have coverage

Topic Discovery Sources

Beyond search data, mine these for high-signal topic ideas:

| Source | What to Look For | |--------|----------------| | Reddit (subreddits in niche) | Recurring questions, complaints, "what I wish I knew" | | YouTube comments on competitor videos | Unanswered questions, complaints about existing content | | Google autocomplete | Long-tail variants of seed keywords | | Answer the Public | Visual map of question + preposition variants | | Client sales calls / CRM | Actual questions customers ask before buying | | Review mining (GMB, Yelp, G2) | Problems solved, language customers use |

Customer language beats algorithmic data — when both sources agree on a topic, it's highest priority.

#content-sop #topic-research #keyword-research #content-planning

View raw JSON →Back to index