The playbook: topical maps and dual optimization.

Start with a topical map, not a keyword list

Most content calendars start with a list of keywords and work outward. That produces fragmented content, overlap, cannibalization, and thin coverage of the actual subject.

Topical mapping inverts the order. Start from the subject you want to be the authority on. Break it into pillars. Break each pillar into clusters. Break each cluster into specific pages. Every page has one job, one primary query, and a clear parent.

For NexcurAI, for example:

• Subject: AI-native consulting
• Pillars (5): Cybersecurity, Web development, Product development, SEO & GEO, Marketing
• Example cluster inside Product development: Evals
• Example pages inside the Evals cluster: Evals primer, LLM-as-judge pitfalls, Regression gates in CI, Scoring rubrics, Eval-first workflows

The map is the single source of truth. Every new piece of content has a slot in the map before it is commissioned. If it does not have a slot, either the map gets extended or the content does not get written.

How to size pillars and clusters

Rule of thumb for a serious topical map:

• 3 to 7 pillars. Fewer and you are underclaiming; more and you are spreading too thin.
• 4 to 10 clusters per pillar. Each cluster should be a sub-topic a specialist would recognize as coherent.
• 5 to 15 pages per cluster at maturity. One pillar page, a handful of cluster pages, and then long-tail detail pages.

At maturity, that is 150 to 1,000 pages. Do not try to ship all of it in month one. A first six months usually covers one pillar fully plus the pillar pages for the rest. A first year covers two to three pillars at depth.

Dual optimization: Google and answer engines

The two targets have different mechanics but overlapping hygiene. What we mean by “dual-optimize” is: every piece satisfies the minimum bar for both.

Google SERP mechanics (still true):

• Match search intent. A “how to” query wants a procedure, not a manifesto.
• On-page basics: title tag, meta description, H1 that matches the intent, internal links.
• Page experience: Core Web Vitals, mobile-first layout, no CLS on headline.
• Links: internal topical links within the cluster, external links to authoritative sources.
• Freshness: dateModified that reflects actual maintenance, not a cron job.

Answer-engine mechanics (from the primer):

• Claim-evidence-source structure with numbers attached.
• Named author, verifiable track record, consistent entity graph.
• Named frameworks and taxonomies with fixed counts.
• First-party experience reports over literature review.
• Clean schema: Article, FAQPage, Organization, Person.

A page that hits both is not longer than a page that hits only one. The discipline costs almost nothing in word count. It costs real effort in editorial rigor.

The monthly content calendar

A sustainable publishing rhythm for a 2 to 4 person content team:

• Pillar essay: 1 per month (deep, 2,500 to 4,000 words, new cluster or major refresh)
• Cluster pages: 2 to 3 per month (medium, 1,200 to 2,000 words, supporting the latest pillar)
• Detail / opinion pages: 4 to 8 per month (short, 700 to 1,200 words)
• Refresh passes: 2 to 4 older pages per month, especially anything with dateModified over 6 months

At that cadence, a pillar fills out in 4 to 6 months. The trap is publishing new content faster than you refresh old content. A 200-page site with 40 pages that have not been touched in two years does not outperform a 150-page site that is all current. Keep the refresh lane staffed.

The production workflow (per page)

One page, end to end, for a dual-optimized piece:

1. Brief (30-60 min). Slot in the map, primary query, secondary queries, target audience, claims we plan to make, evidence we plan to cite, desired framework name or structure, author.
2. Research (1-3 hrs). Gather evidence. Check what is already ranking and already being cited. Identify the gap.
3. Draft (2-6 hrs). Claude-assisted, human-directed. Strong opening claim. Structure by section. Named framework if applicable. Concrete numbers.
4. Edit (1-2 hrs). Editorial pass by a named human. Trim. Sharpen. Add schema. Verify facts.
5. QA (30 min). Lighthouse. Schema validator. Internal link check. Table of contents if over 1,500 words.
6. Ship. Publish, submit sitemap, post to newsletter if applicable, cross-link from siblings in the cluster.
7. Measure. Google Search Console rankings + share-of-answer in the 4 engines after 30 and 60 days.

Measurement: what to track monthly

Four dashboards, reviewed monthly:

• Organic traffic. Search Console: clicks, impressions, CTR, average position. Segment by pillar.
• Share of answer. Citation rate per engine across your ICP query set. Aim for “up and to the right” month over month, not absolute targets in month one.
• Conversions. Which pages drive contact, signup, or whatever your conversion is. A page that ranks but does not convert is not done.
• Freshness. Count of pages with dateModified older than 6 months. This should trend down, not up.

Do not build vanity dashboards. A one-page monthly report with those four numbers, trend arrows, and a short narrative is enough.

Topical authority signals we watch

Beyond rankings and citations, authority shows up as:

• Branded search volume for your name plus the subject (“[Brand] evals”, “[Brand] pentest”). If this is climbing, you are being associated with the topic.
• Direct traffic to deep pages. People bookmarking and returning.
• Referral links from peers - other operators linking to your piece as the reference. This is the slowest to build and the most durable.
• Podcast and newsletter mentions, which often do not show as traceable referrers but show as brand-search spikes.

Anti-patterns that kill playbooks

• Hiring a generalist content writer with no operator experience. The voice goes corporate, the evidence gets generic, nothing gets cited. Hire editorially-strong operators who can Claude-assist, not content-mill writers.
• AI-slop bulk publishing. Google and answer engines both downweight it. Fast. Do not pile up unedited output for volume.
• Chasing every keyword. Off-topic ranking distracts from the topical authority you are trying to build. Say no to keywords that do not fit the map.
• Not shipping the pillar page. Cluster pages with no pillar to link up to are orphans. Every cluster has a pillar.
• Confusing thought leadership with GEO content. A manifesto about the future of X does not answer the question “how do I X.” Both have their place; know which you are writing.
• Publishing and forgetting. The refresh lane is not optional. Set a calendar reminder.

Team shape

Minimum viable team:

• One editor / head of content. Owns the map, the calendar, the tone. Claude-fluent, operator mindset.
• One or two contributors. Write drafts, maintain pages. Often subject-matter specialists who learn to Claude-assist.
• One technical owner. Schema, performance, analytics, Search Console. Can be part-time.

You do not need a content team of six. You need two good people with a clear map and a discipline.

When to talk to us

If you are trying to build this in-house, our SEO & GEO engagements deliver the topical map, the initial 30 to 90 pages, and the monthly operating rhythm - then we hand it off. Start a conversation if that fits.