Topical Authority Map

Ranking for a competitive head term is rarely a page-level problem. Google evaluates whether a site is a credible source on a subject, and one good page inside an unrelated site doesn't establish that.

The map is the answer to a specific question: which set of pages, published in which order, makes this site the obvious source on this topic?

Step 1: Define the topic honestly and narrowly

The most common failure is picking a topic the site has no business owning. A four-person startup cannot own "marketing". It can own "cold email deliverability for B2B SaaS".

Narrow until all three are true:

• The site has a genuine claim. Real expertise, real customers, real data in this area.
• The topic is small enough to cover completely in 15 to 40 pages, not 400.
• It connects to what's sold. Authority on a topic nobody pays for is a hobby.

Write the topic as one sentence, then write the sentence a reader would use to describe what the site is for. If those two don't overlap, adjust.

Step 2: Map the subject, not the keywords

Start from the subject's actual structure rather than from a keyword export. The map should read as a curriculum.

For any topic, enumerate:

• Concepts — what things are, and what the terms mean
• Processes — how things are done, step by step
• Decisions — how to choose between options
• Problems — what goes wrong and how to fix it
• Comparisons — this versus that, alternatives to this
• Evidence — benchmarks, data, results, case studies
• Edges — when the usual advice doesn't apply

A map missing an entire category is why sites plateau. Most blogs are 90% concepts and processes with no decisions, problems or comparisons, which is exactly where the commercial intent and the AI citations are.

Then check the coverage against reality: People Also Ask, related searches, the questions in support tickets and sales calls, and the subheadings competitors use. Anything asked repeatedly that the map doesn't answer is a gap.

Step 3: Structure it as hubs and clusters

PILLAR: Cold email deliverability                    ← head term, the hub
│
├── CLUSTER: Authentication
│    ├── What is SPF and how to set it up
│    ├── What is DKIM
│    ├── What is DMARC and which policy to use
│    └── How to check if authentication is passing
│
├── CLUSTER: Domain and inbox setup
│    ├── How to warm a new sending domain
│    ├── Should you use a separate sending domain
│    └── How many emails per inbox per day
│
├── CLUSTER: Diagnosing problems
│    ├── Why emails land in spam
│    ├── How to check if a domain is blacklisted
│    └── What to do after a deliverability collapse
│
└── CLUSTER: Choosing tools
     ├── Best cold email tools compared
     ├── Tool X vs Tool Y
     └── Deliverability testing tools

• The pillar targets the head term and links down to every page in the map. It's a real page a person would read, not a link directory.
• Each spoke links up to the pillar and sideways to two or three genuine siblings.
• One page per question a reader would ask separately. Not one page per keyword variation.
• Clusters are groups of pages, not a page type. A cluster with one page in it isn't a cluster; merge it.

The linking implementation matters as much as the map. See internal-linking-architecture.

Step 4: Order the publishing

The order is a strategy decision, not an editorial one.

Publish the commercial and comparison pages early. They convert, they're usually less competitive than the head term, and they're what AI answer engines cite when someone asks what to use. The instinct to "build authority first with informational content, then monetise" delays revenue by a year and often never arrives.

A workable sequence:

1. Two or three comparison or decision pages. High intent, cited by assistants, immediate business value.
2. The problem pages. People searching a problem are ready to act and these are usually under-served.
3. The core concept pages the other pages need to link to.
4. The pillar, once there's enough beneath it to make it a genuine hub. A pillar published first is a page linking to nothing.
5. The long tail, ongoing.

Publish in batches of five or six, then measure before continuing. A map planned in full and published in full without measurement is a twelve-month bet on assumptions.

Step 5: Know when it's working

Signals, in the order they appear:

1. Long-tail pages start ranking within weeks. The easiest wins arrive first.
2. Pages start ranking for terms you didn't target. This is the real signal that the site is being read as a source on the subject.
3. New posts in the cluster rank faster than earlier ones did. Compounding.
4. The pillar starts moving on the head term. Slowest, and it's the outcome.
5. The brand starts appearing in AI answers for questions in this space, because there's a coherent body of work to synthesise from.

If long-tail pages aren't ranking after two months, the problem is upstream — the topic is too competitive for the site, or the pages aren't better than what ranks. Don't publish more into a map that isn't converting; diagnose first.

Don't do these

• Don't map 200 pages. Nobody executes a 200-page plan. Map 20, ship them, extend.
• Don't split one question across three pages to hit a page count. That's cannibalisation with extra steps. See keyword-cannibalization.
• Don't publish the pillar first. It needs the cluster to hub.
• Don't build a map from a keyword export alone. Exports are full of variations of the same question and missing the questions nobody has written about yet, which are the valuable ones.
• Don't cover a topic the site can't credibly claim. Coverage without expertise is exactly what quality assessment is designed to catch.
• Don't confuse coverage with completeness. Twenty thin pages don't establish authority. Eight genuinely good ones do.

The output

# Topical map: [topic]

**Claim:** why this site can credibly own this
**Scope:** what's in, what's deliberately out
**Head term:** [term] — the pillar target

## Pillar
[Title] — /slug — targets [term]

## Cluster: [name]
| Page | Slug | Query | Intent | Priority | Status |
|---|---|---|---|---|---|

## Publishing order
1-6:  [pages] — first batch, measure after
7-12: [pages]

## Gaps needing input from the business
- [Data, examples, expertise required and from whom]

Where the data comes from

Two inputs decide whether the map is right, and both are hard to get by hand: which questions in this space have real demand, and which of them competitors already own. Getting these wrong means executing a well-structured plan for the wrong twenty pages.

Search Console shows the terms the site already gets impressions for, which is the best free starting point for finding the topic you have a claim to. Keupera covers the rest with topic silos, question mining, gap analysis and right-to-win scoring, plus prompt research for the questions people ask LLMs rather than search engines, which is where the decision and comparison pages come from. Over MCP:

claude mcp add --transport http keupera https://mcp.keupera.com/mcp -s user

Related skills

keyword-research-briefs, internal-linking-architecture, keyword-cannibalization, comparison-alternatives-pages, glossary-definition-hub, content-refresh