The 90-minute topical map: an operator's recipe.

A topical map is the spine of a ranking strategy.

It is not a keyword list. It is not a content calendar. It is a structured tree of all the subtopics that belong under one pillar, ordered by buyer urgency, with a clear publishing sequence. Without it, content programs ship random posts that compete with each other. With it, they ship clusters that compound.

The standard advice says: buy Ahrefs, pull 2,000 keywords, run them through clustering tools, refine for weeks. That works. It also takes weeks. The 90-minute version uses sales calls instead of keyword data and produces a draft that is roughly 80% as good — which is enough to start shipping. Volume research validates the map later.

Here is the recipe.

What you need

Five sales call transcripts. That is the whole input.

If you do not have sales calls, substitute: customer support transcripts, recorded product demos with prospects, community forum threads with active buyers, or notes from 5 customer interviews. The principle is the same — real language from real buyers.

Tools: a whiteboard or a Miro board. No SaaS required.

Stage 1: Gather (10 minutes)

Pull 5 transcripts. Skim each for the section where the prospect is articulating their problem before you started pitching. The first 10 to 15 minutes of a discovery call is gold; the demo part is mostly noise.

If transcripts are not available, use call recordings and skim through the prospect’s monologue sections. Five calls is enough; ten is overkill for a first draft.

Stage 2: Extract (30 minutes)

For each transcript, list every named entity the prospect mentions. Aim wide. Pull:

• Named tools, platforms, products
• Named methodologies, frameworks, approaches
• Named competitors, alternatives, peers
• Named pain points (specific phrasing they use)
• Named outcomes they want
• Named processes they are trying to fix
• Named stakeholders involved in the buying decision

Do not edit yet. List everything. A 30-minute call usually produces 20 to 40 named entities. Five calls produces 80 to 150 raw entities. About 20 to 30% will be duplicates across calls — that is fine, the duplicates are signal.

Stage 3: Cluster (20 minutes)

Group the entities into 3 to 6 clusters. The clusters are not predefined. Let the entities tell you what the natural groupings are.

For a B2B SaaS selling sales tooling, a typical 5-cluster shape:

• Cluster A: Pipeline visibility entities (forecasting, deal stages, win rates)
• Cluster B: Rep enablement entities (call coaching, playbooks, training)
• Cluster C: Process entities (handoffs, qualification frameworks, account scoring)
• Cluster D: Adjacent tool entities (CRM, conversation intelligence, scheduling)
• Cluster E: Outcomes entities (productivity, ramp time, quota attainment)

Each cluster will contain 10 to 20 entities. The clusters become your eventual content sub-pillars; the entities inside become individual posts.

If you have more than 6 clusters, you are probably looking at multiple pillars (a category, not a topic). If you have fewer than 3, the topic is too narrow for a real content program.

Stage 4: Rank by buyer urgency (15 minutes)

For each cluster, rank by how often buyers articulate urgency around the topic. Three tiers.

• Tier 1. Mentioned by 4 to 5 of the 5 prospects as a current pain. High urgency.
• Tier 2. Mentioned by 2 to 3 prospects as a current pain or by 4 to 5 as a future concern. Medium urgency.
• Tier 3. Mentioned by 0 to 1 prospects. Low urgency but may still belong on the map.

Tier 1 entities are the publishing priority. They map to high-intent buyer queries that close deals.

This is where the recipe diverges most sharply from keyword-volume-driven content strategy. Volume tells you what gets searched. Urgency tells you what gets bought. For B2B, urgency wins.

Stage 5: Sketch the publishing order (10 minutes)

Two principles.

Cluster completion beats query coverage. Pick one cluster. Plan to ship 4 to 8 posts that cover every Tier 1 entity in that cluster before you ship anything in another cluster. Authority compounds when clusters are filled; it does not compound when posts are scattered across topics.

Pillar first, then spokes. Within a cluster, ship the pillar page (the broadest entity) before the spokes (the specific entities). Spokes link up to the pillar; the pillar absorbs link equity from the cluster. See pillar pages for the structure.

For the example clusters above, the first 8 posts might be:

1. Pillar: pipeline visibility for B2B sales (Cluster A pillar)
2. Spoke: forecasting accuracy frameworks (Cluster A)
3. Spoke: deal stage hygiene (Cluster A)
4. Spoke: win-rate diagnosis (Cluster A)
5. Spoke: pipeline coverage ratios (Cluster A)
6. Pillar: rep enablement at scale (Cluster B pillar)
7. Spoke: call coaching frameworks (Cluster B)
8. Spoke: playbook adoption (Cluster B)

Cluster A is complete before Cluster B starts. Cluster B is complete before Cluster C starts. This is the cluster-completion principle in action. See topical authority for the underlying theory.

Stage 6: Validate with volumes (later)

After the draft is in front of you, run the entities through whichever keyword tool you have. Tag each with a search volume, a keyword difficulty, and an intent classification.

Use the data to:

• Identify entities with zero search volume that should be moved off the map (rare — usually you keep them as supporting content for the cluster)
• Sequence within a tier (within Tier 1, ship the higher-volume entities first)
• Flag entities with extreme difficulty that need a different long-tail angle

The validation step takes about 2 hours when you have the data and the toolkit. It is separate from the 90-minute draft. The point is that you do not need the data to start; you need it to refine. See competitive keyword research for the volume-research workflow.

What you should not include

A few patterns that bloat topical maps without earning their slot.

Generic SEO entities. “How to do SEO”, “SEO best practices”, “SEO checklist”. These have huge volume and zero pipeline contribution. They belong on a category map, not your topical map.

Aspirational entities. Topics you wish your prospects cared about but they do not actually mention. If it does not show up in the calls, it does not belong on the v1 map. You can add it in v3.

Entities outside your buyer’s vocabulary. If you cluster 40 entities and you find yourself adding 10 more from your own knowledge of the category, stop. The map is for the buyer, not for you. Add the missing entities to a research backlog and revisit.

What this is not

Two clarifications.

The 90-minute map is a v1 draft. It is good enough to start publishing. It is not a permanent strategy. Revisit quarterly with fresh sales calls and update.

This is not a substitute for keyword research. The map tells you what to publish. Keyword research within each entity tells you how to title each post, what variants to target, what intent to satisfy. Both matter; the map matters first because it provides the structure keyword research operates within.

A worked example

For a Series B B2B SaaS I worked with in late 2025, the 90-minute draft produced:

• 5 clusters
• 47 named entities total
• 14 entities tagged Tier 1
• A clear publishing order through the first 4 weeks

After validating with Ahrefs:

• 41 entities had measurable search volume (mostly 100 to 1,000 SV)
• 4 had volume too low to bother with on their own (kept as supporting content)
• 2 we dropped because difficulty was unrealistic

We shipped the first 8 posts in 9 weeks. By week 12, the site ranked top-3 for 11 of the 14 Tier 1 entities. The cluster-completion sequencing did most of the work.

This is a normal outcome, not exceptional. Most B2B sites that ship cluster-complete content with answer-first openings (see how to get cited by ChatGPT) see similar lift within 8 to 14 weeks. The map is what makes the sequencing possible. Without it, content programs ship one post at a time in random order and authority never compounds.

What to do tomorrow

The smallest version of this work.

1. Block 90 minutes on the calendar.
2. Pull 5 sales call transcripts (or substitute: 5 customer interviews, 5 demo recordings, 5 support tickets from active buyers).
3. Run the recipe above.
4. Output: one page with 5 clusters, 30 to 60 entities, a Tier 1/2/3 ranking, and a publishing order for the first 8 posts.

If the output looks too small, your topic is probably too narrow. If it looks unwieldy, your topic is probably a category. Both are fine — adjust scope and run again.

The map is the spine. Without it, every other piece of content strategy work is decoration. With it, the rest of the program has something to compound against.