When people hear "topical map," they picture a spreadsheet with 500 keywords sorted by volume. That's not a map — that's a dump. A real topical map is a page-by-page architecture: every page has a name, a job, one search intent, a place in the hierarchy, and a list of pages it links to. For SW Pavers, a paver and hardscape contractor, that architecture came out to exactly 149 pages. Here's the process that produced it.
Step 1 — Map the business before the keywords
The first pass has nothing to do with search tools. I sit with what the company actually sells and where the money is: paver installation, driveways, pool decks, patios, walkways, retaining walls, repairs, sealing, and the service area it can realistically win. If a page can't connect to a job the crew actually books, it doesn't earn a slot on the map — traffic without revenue is decoration.
Step 2 — Mine the entities, not just the queries
Then the research pass: Ahrefs, competitor sitemaps, Google's own suggestions, People Also Ask, and — underrated — the client's sales conversations. The goal isn't a keyword list, it's the entity set of the topic: materials (travertine, brick, concrete pavers), applications (driveway, pool deck, lanai), problems (sinking, weeds, drainage, cracked slabs), decisions (pavers vs. stamped concrete, cost questions, permits), and maintenance (sealing, cleaning, repairs). Cover the entities and the long-tail queries come with them — thousands of phrasings collapse into a few dozen real topics.
Step 3 — Split by intent, ruthlessly
Every topic then gets split by what the searcher is actually trying to do, because intent decides the page type:
- Hire intent → service pages and location pages. Commercial copy, quotes, proof, schema.
- Compare intent → "pavers vs. X" pages. Honest comparisons that pre-sell the estimate.
- Cost intent → pricing guides. The pages competitors are scared to write.
- Learn / fix intent → blog and guide content that catches homeowners early and routes them to the service page.
One intent per page is the rule that makes the whole thing work. The moment a page tries to serve two intents, it ranks for neither — and it starts competing with its own siblings.
Step 4 — The anti-cannibalization pass
Before anything gets a green light, every planned page is checked against every other planned page and everything already indexed: does another URL already target this intent? Could Google reasonably confuse the two? On a 149-page map there are thousands of potential collisions, and this pass is exactly why the map is written down before the content is. Merges happen on paper, not as redirects after six months of confusion.
The map is the strategy. The content is just the map being executed.
Step 5 — Hubs, spokes, and the internal-link plan
The pages arrange into hubs and spokes: a pillar per money topic (driveway pavers), spokes covering its entities and questions (cost, materials, drainage, permits, repair), every spoke linking up to its pillar and sideways to its siblings, every pillar linking down. The internal links are specified in the map itself — anchor text included — so authority flows where it's pointed, not wherever the writer happened to remember a link.
Step 6 — Build order: money first, authority behind it
149 pages don't ship at once. Service and location pages go first — they can convert from day one, even at partial rankings. Then comparison and cost pages, because they're closest to a signed contract. Informational spokes fill in behind, each one making the pillar above it stronger. Every page ships with its schema, its meta, and its internal links already specified — staged as a draft, checked, then published.
What the data did next
The honest part: topical authority compounds slowly, then suddenly. Five months into the build-out, the SW Pavers property shows 419 clicks and 69.7K impressions in Search Console with the impression curve still bending upward as the map indexes — you can see the unedited screenshot on the SEO page. Impressions lead, clicks follow, and every new spoke that indexes makes the older pages a little heavier.