A Real B2B Topic Cluster Example: How One Seed Keyword Became 14 Pages

Most “B2B topic cluster examples” online are abstract — a CRM diagram, a SaaS blog, a tidy hub-and-spoke drawing with no real data behind it. This is different. This is a real B2B topic cluster example built from a single seed keyword for an industrial manufacturer, with every decision shown — including the two mistakes the process caught and fixed before a single page was written.

The short version, up front: one seed keyword — patch antenna — expanded into roughly 250 candidate terms, was filtered down to about 50 real seed keywords, and after validating intent against live Google results, resolved into a 14-page topic cluster: 1 pillar, 1 commercial page, 3 product/application pages, and 9 supporting articles. Everything below is the exact, repeatable path from one to fourteen.

By the end you’ll have a process you can run on your own seed keyword this week, plus the free template we used. One line to anchor it all: a topic cluster is one pillar page plus supporting pages that all target a single theme and link to each other — and the right number of pages is decided by search intent, not by a word-count target.

The 5 steps at a glance

1. Filter competitor categories into in-scope seed keywords.
2. Expand the seed — then filter by business scope a second time.
3. Validate intent on the SERP: merge, split, or fold.
4. Prioritize every page on a value × winnability matrix.
5. Map it into one keyword map and build pillar-last.

Now the same five steps on a real product, with the real numbers.

The setup: one seed keyword, one real B2B product

The site is an industrial RF/antenna manufacturer, and its business scope is narrow on purpose: GNSS/GPS and cellular (LTE/4G/5G) antennas — not WiFi, not Bluetooth, not LoRa or RFID.

Write your business scope down before you open a single tool. It matters more than any keyword metric, because it silently decides which terms deserve a page and which are noise. Half the discipline in this whole example is just refusing to drift outside that one line.

We started with one specific, product-level seed — patch antenna — not a vague head term like “antennas” or “RF.” For a new B2B site, a narrow seed you can realistically win beats a broad head term you can’t.

Step 1 — Turn competitor categories into seed keywords (not all of them)

We pulled the full product-category lists from six authoritative competitor sites — about 250 terms — and filtered them. The most common mistake here is treating every competitor term as a keyword. Most of them aren’t.

Those 250 terms fall into eight types, and only three of them survive:

• Keep — real in-scope product categories: gps patch antenna, ceramic patch antenna, rhcp patch antenna.
• Combine — technology/form modifiers that only work attached to a product noun: 5g → 5g patch antenna, ceramic → ceramic patch antenna.
• Drop — competitor brand/series names (Thunder, Raptor Max, LoPro), navigation labels (“Products – All”, “Category Overview”), cross-line products (speakers, transformers), and anything out of scope.

Figure 1 — ~250 competitor terms filtered down to ~50 seeds. (alt: “funnel filtering competitor antenna terms into keep, combine, drop, down to ~50 seed keywords”)

flowchart LR
    A["~250 competitor terms<br/>(6 vendor sites)"] --> F{"Filter by<br/>business scope"}
    F -->|Keep| K["gps patch antenna<br/>ceramic patch antenna<br/>rhcp patch antenna"]
    F -->|Combine| C["5g → 5g patch antenna<br/>active → active gps patch"]
    F -->|Drop| D["brand names · nav labels<br/>WiFi / Bluetooth (out of scope)"]
    K --> S["~50 real seed keywords"]
    C --> S

After filtering, ~250 raw terms became ~50 real seed keywords, and brand names plus out-of-scope technologies were already cut. The takeaway: a competitor’s navigation menu is a starting point, not a keyword list — filter it against your scope before anything else.

(Full method: How to pull keywords from competitor websites the right way.)

Step 2 — Expand the seed, then filter again (the scope leak)

Next we expanded patch antenna with Google autocomplete, People Also Ask, related searches, and a keyword tool. Here’s a trap almost nobody warns you about: expansion re-introduces out-of-scope terms your clean seed list never had.

The expansion happily surfaced patch antenna for wifi, bluetooth patch antenna, 5 ghz wifi patch antenna, and patch antenna for 2.4 ghz. Worse, an AI classification pass rated several of them “medium priority.” They look great on paper — real volume, obviously related.

But this manufacturer doesn’t make WiFi or Bluetooth antennas. Those pages would describe products that don’t exist, pull in unqualified traffic, and blur the site’s topical focus. So we ran the business-scope filter a second time — on the expansion — and dropped every out-of-scope term.

The rule that saves whole clusters: filter by business scope twice — once on the seeds, and again after expansion. Expansion tools optimize for “related,” not for “things you actually sell,” and neither do AI suggestions — they don’t know your scope unless you enforce it.

Step 3 — Let the SERP decide: merge, split, or fold

Before assigning any keyword to a page, we checked Google’s top 10 for it. The SERP — not your opinion — tells you whether two keywords are one page or two, and whether a keyword deserves a page at all. Three concrete decisions came out of it.

Merge — when the results overlap. patch antenna manufacturer and patch antenna supplier shared roughly 8 of their top 10 URLs — the same directories (Metoree, ThomasNet, everythingRF) and the same vendors (Pasternack, Suntsu). Identical SERP means identical intent, so these become one commercial page, one term primary and the other secondary — never two pages cannibalizing each other.

Fold — when the SERP is unwinnable. patch antenna beamwidth returned almost nothing but IEEE papers, arXiv PDFs, ResearchGate threads, and engineering forums — zero buyer intent and unwinnable for a young site. So instead of a doomed page, we folded it in as an H2 inside the radiation-pattern article.

Confirm the page type — by who already ranks. gps patch antenna was dominated by product and collection pages (gnss.store, SparkFun, Taoglas), with only a couple of application-note PDFs mixed in. That makes it a product page, not a blog post.

The whole skill in one sentence: SERP overlap decides merge vs. split, and the pages that already rank decide your page type.

(Full method: How to validate search intent with the SERP.)

Step 4 — Prioritize with a value × winnability matrix

We did not score keywords across eight weighted dimensions chasing a magic total — for a new site, that’s analysis paralysis. Instead, every page goes on a simple 2×2: business value versus how winnable it is for a site with little authority.

	High value	Low value
Winnable now	P0 — write first	P1 — long-tail support
Hard now	Defer (build, rank later)	Drop

Figure 4 — Each page placed by value × winnability. (alt: “keyword priority matrix: value vs winnability with P0, P1, Defer, Drop quadrants”)

quadrantChart
    title Keyword priority: value vs winnability
    x-axis "Low winnability" --> "High winnability"
    y-axis "Low value" --> "High value"
    quadrant-1 "P0 — write first"
    quadrant-2 "Defer — build, rank later"
    quadrant-3 "Drop"
    quadrant-4 "P1 — long-tail support"
    "gps patch antenna": [0.66, 0.86]
    "manufacturer": [0.56, 0.90]
    "patch antenna (head)": [0.16, 0.80]
    "how to choose": [0.72, 0.42]
    "beamwidth (fold)": [0.22, 0.26]
    "patch antenna for wifi": [0.30, 0.10]

This is exactly why the head term patch antenna is Defer, not a day-one target. It’s valuable, but a brand-new site won’t outrank the incumbents for it early. So we build it as the pillar and let authority accrue, while the winnable product and long-tail pages bring the first traffic and inquiries. You don’t earn the head term by targeting it — you earn it by winning everything around it first.

(Full method: The keyword priority matrix — value × winnability.)

The result: one seed keyword, fourteen pages

Here’s the full cluster the single seed produced:

Role	Pages	Priority
Pillar (hub)	patch antennas	Defer
Commercial	patch antenna manufacturer (+ supplier)	P0
Product / application	gps patch antenna · rhcp patch antenna · patch antenna for drones	P0–P1
Supporting blogs	design · gain · bandwidth · radiation pattern (incl. beamwidth) · advantages & disadvantages · how to choose · rhcp vs lhcp	P1–P2

Figure 5 — The 14-page cluster: pillar hub, spokes, and the conversion funnel into the money pages. (alt: “14-page B2B topic cluster hub-and-spoke diagram linking supporting pages to the pillar”)

flowchart TB
    PI["Pillar: patch antennas"]
    PI --- C1["Commercial: manufacturer"]
    PI --- C2["Product: gps patch antenna"]
    PI --- C3["Product: rhcp patch antenna"]
    PI --- A1["App: for drones"]
    PI --- A2["App: applications"]
    PI --- B1["Blog: design"]
    PI --- B2["Blog: gain"]
    PI --- B3["Blog: bandwidth"]
    PI --- B4["Blog: how to choose"]
    PI --- B5["Blog: rhcp vs lhcp"]
    PI --- B6["Blog: pros & cons"]
    PI --- B7["Blog: radiation pattern"]
    PI --- B8["Blog: basics"]
    B4 --> C2
    B5 --> C3
    A1 --> C2
    C2 --> C1
    C3 --> C1

One seed keyword. Fourteen pages. Each maps to exactly one search intent, and every page links back to the pillar — that internal-linking structure is what signals real topic depth to Google, instead of one thin post floating alone.

(See exactly how the map is assembled: Keyword mapping for SEO — with a free template.)

What this method caught that most keyword research misses

A keyword-list-and-go approach would have shipped three real problems:

1. The WiFi/Bluetooth scope leak — pages for products the company doesn’t sell, dragging in unqualified visitors and diluting topical authority.
2. A wasted “beamwidth” page — a doomed attempt to outrank IEEE for a query no buyer types.
3. Two cannibalizing money pages — “manufacturer” and “supplier” fighting each other in Google instead of being one strong page.

Transparency is the point: a real method shows its corrections, not just a clean final diagram. That honesty is what earns trust — from readers deciding whether to copy your process, and from AI search engines deciding whose content to cite.

How long does this take — and when will it rank?

Building the map for one seed is a focused half-day once you’ve done it twice. Writing the 14 pages is the real work, and you don’t write them in the cluster’s reading order — you build the winnable pages first.

Set expectations honestly: on a newer site, a long-tail supporting page often shows movement in 4–8 weeks, while the competitive pillar and commercial pages can take months. That’s not failure; it’s the plan. Win the long-tail first, and the head terms follow as topical authority compounds.

How to replicate this (free template)

Run the same five steps on your own seed keyword:

1. Filter competitor categories into in-scope seeds.
2. Expand the seed — then filter by scope again.
3. Validate intent on the SERP: merge, split, or fold.
4. Prioritize each page on the value × winnability matrix.
5. Map it all into one keyword map and build pillar-last.

Grab the keyword map template and priority matrix we used for this exact cluster — download both here — and turn your first seed keyword into a real cluster.

Or skip the manual work — download the skill pack. The four agent skills behind this exact workflow — serp-intent-validator, keyword-map-builder, content-brief-writer, and content-production-planner — are packaged for Claude Code, Codex, and Claude.ai: Download the SEO skill pack (.zip). Drop them into your agent and run this whole method on your own seed keyword.

FAQ

How many pages should one keyword become? As many as it has distinct search intents — here, one seed produced 14. Let the SERP, not a quota, decide where to merge and where to split.

How do you decide whether to merge or split keywords into pages? Compare the top 10 Google results for each. Three or more shared URLs means the same intent — combine them into one page. Clearly different results mean separate pages.

Does this work for industrial or manufacturing B2B, not just SaaS? Yes — this entire example is a physical industrial product line (GNSS and RF antennas), not a software topic. The method is industry-agnostic.

What if a keyword has search volume but is out of my business scope? Drop it. A page for a product you don’t sell attracts the wrong audience and weakens your topical authority — volume doesn’t change that.

How long until these pages rank? For a newer site, expect 4–8 weeks before a supporting page moves, and longer for the pillar. Win the long-tail product and supporting pages first; the head terms follow.

---

By Taylor Yang — written from a real cluster build. Last updated 2026-06-03.