What Silo-and-Cluster Content Actually Means
Strip the jargon and it is simple: a silo is one core service (say, roof replacement, or drain cleaning, or panel upgrades) and a cluster is the group of supporting articles that surround it. The silo page targets the money keyword: broad, commercial intent, the page you'd send a bid on. The cluster pages target the specific questions a homeowner types into Google before they ever call a contractor: cost breakdowns, comparisons, timelines, warranty questions, seasonal problems, material choices.
Every cluster page links up to the silo page. The silo page links down to its clusters. Related clusters link sideways to each other where the topic genuinely overlaps. That internal linking is not decoration, it is the mechanism. Search engines and AI crawlers use those links to figure out which pages belong to the same subject and which page is the authority the others defer to. A site with 20 disconnected blog posts about roofing has 20 orphans. A site with the same 20 posts wired into a roofing silo has one authority with 20 supporting proofs.
The distinction matters more now than it did five years ago because AI answer engines (ChatGPT, Gemini, Perplexity, Google's AI Overviews) don't cite the loudest page, they cite the page that reads like it belongs to a complete, coherent body of work on the topic. A silo is what makes that judgment easy for the machine to make in your favor. Think of it the way a homeowner sizes up a job site: a truck with matching signage, matching uniforms, and a crew that answers every question the same way reads as a real operation. A truck with a magnet sign and one guy who mumbles reads as a gamble. Search engines and AI models are making the same read on your website.
None of this replaces the mechanics of ranking: the keyword targeting, the technical setup, the backlinks. That work lives in the SEO layer and gets built on top of the content, it doesn't substitute for it. Silo-and-cluster is the architecture the content lives inside; it's the difference between a stack of lumber and a framed wall.
Why a Single Blog Post Never Outranks a Silo
Contractors who tried content marketing once and quit usually tried it as a series of one-off posts: a holiday post, a "why choose us" post, maybe a seasonal maintenance tip. Each post stood alone, linked to nothing, and competed against pages built by sites that had 15 or 20 interlinked articles on the exact same sub-topic. That's not a fair fight, and it's not really the same game.
A single post has to win on its own merits: word count, keyword match, backlinks, page experience. A silo page gets to borrow authority from every cluster page pointing at it. When a homeowner searches "how much does a new roof cost in [region]" and lands on a cluster page that then links to your roofing service page, and that same cluster page also gets cited when they ask an AI assistant the same question, the silo is doing double duty: it earns the click and it earns the citation.
The other reason single posts fail is coverage. Google and AI models both reward completeness on a topic because incomplete coverage is a signal of a thin, low-effort site. If a competitor's roofing silo answers cost, materials, timeline, insurance claims, storm damage, warranty length, and financing, and yours answers only "why choose our roofing company," the machine has no reason to treat your page as the authoritative one, no matter how well that single page is written.
- One post: competes alone against a competitor's full cluster
- Silo of 12-25 pages: pools relevance and link equity behind the money page
- Cluster pages: catch long-tail questions the silo page is too broad to rank for
- Cross-linking: tells the crawler (and the AI model) these pages are one body of work
This is also why a stale blog with three posts from 2022 doesn't help you. Coverage that stopped isn't coverage, it's a snapshot of what you knew about the topic three years ago.
How Many Cluster Pages Does a Trade Actually Need
There's no universal number, but there is a working range: 94+ cluster pages is typical for a fully built-out multi-service contractor site covering several silos (roofing, gutters, siding, for example), which breaks down to somewhere between 12 and 25 pages per individual silo depending on how many real sub-questions that service generates. A niche trade like pressure washing might top out around 12 solid clusters. A trade with heavy regulatory, seasonal, and cost-comparison angles like HVAC or roofing can support 25 or more without padding.
The right number is determined by real customer questions, not a quota. Pull from three sources: what customers actually ask on the phone before booking, what shows up in Google's "People also ask" boxes for your core service terms, and what your competitors' sites are missing. If you run out of genuine questions before you hit 12, you've probably got a niche enough service that a smaller, tighter cluster is the honest answer, and padding it with filler articles does more harm than good, both to readers and to how an AI model judges your site's substance.
A good gut check before committing to a number: list every question a new customer asked on the phone in the last month, then cross out the ones your homepage already answers. What's left is your real cluster list, and it usually lands closer to reality than a generic keyword tool's suggestion, because it came from an actual customer instead of a spreadsheet.
| Trade pattern | Typical cluster size | Why |
|---|---|---|
| Single-service niche (pressure washing, gutter cleaning) | 10-14 pages | Fewer real sub-questions, less seasonal/regulatory variation |
| Mid-complexity trade (plumbing, electrical, landscaping) | 15-20 pages | Emergency vs. planned work, permit questions, multiple service lines |
| High-complexity trade (roofing, HVAC, remodeling) | 20-25+ pages | Insurance claims, financing, material comparisons, code/permit variance, seasonal urgency |
Building the full number in one sprint is neither necessary nor smart. A realistic build-out runs several months, publishing on a steady cadence so the silo grows the way a real, ongoing authority would, not the way a content farm dumps 25 pages in a week and goes quiet.
What Order to Write the Pages In
Start with the silo page itself. It has to exist and be solid before cluster pages have anywhere to point. That's your core service page: what the service is, what it costs in ranges, how long it takes, who it's for, who it isn't for. If that page is thin or missing, publishing cluster content around it is building a roof with no walls under it.
Next, write the two or three cluster pages that answer the highest-intent questions: usually cost ("how much does X cost") and a direct comparison ("X vs Y") or a warning post ("signs you need X"). These pages tend to carry the most searches and the most AI-answer visibility because they match exactly how people phrase real questions, whether they're typing into Google or asking an assistant out loud.
From there, fill in supporting angles in roughly this order:
- Cost and pricing breakdowns (the single highest-demand cluster type across every trade)
- Comparison posts (material vs. material, method vs. method, DIY vs. professional)
- Problem/symptom posts (signs of failure, when to call, what happens if you wait)
- Process and timeline posts (what to expect, how long it takes, permit and inspection steps)
- Seasonal and regional posts (climate-specific wear, local code quirks, storm-season urgency)
- Trust and differentiation posts (warranty specifics, financing, what's included vs. not)
Publishing in this order means the pages most likely to earn traffic and links go live first, and every page after that has an established silo to strengthen rather than a silo it's waiting to help build. It also means an AI model crawling your site mid-build still sees a coherent, high-value cluster rather than a scatter of half-finished topics.
One more practical note on sequencing: publish on a visible, steady cadence, whether that's weekly or biweekly, rather than in bursts followed by long silences. A silo that grows steadily reads as an active, maintained authority. A silo that jumps from zero to 20 pages overnight and then goes quiet for a year reads the way it actually is: a project someone abandoned.
The Internal Linking Rules That Make a Silo Work
Content without linking discipline is not a silo, it's a pile. Three rules cover almost all of it. First, every cluster page links up to its silo page, usually once in the body copy and once in a closing call-to-action, using anchor text that describes the service rather than a bare "click here." Second, the silo page links down to its most important clusters, typically the cost page and the comparison page, since those are the pages doing the heaviest lifting for both rankings and AI citations. Third, cluster pages link sideways to each other only when the connection is real. A roofing cost page can honestly link to a roofing materials comparison page. It should not force a link to an unrelated gutter-cleaning post just to pad link count.
What kills a silo's authority is linking across silos indiscriminately, treating every page on the site as one giant undifferentiated web. If your roofing content links heavily into your landscaping content and vice versa, you've diluted the topical signal each silo was supposed to build. Keep the cross-links inside the trade they belong to, and let the site's main navigation (not individual body links) handle getting a reader from one silo to another.
The other detail that gets missed: a link only counts as reinforcing the silo if the anchor text and surrounding sentence actually describe the destination page. A cluster page that links to the silo page with the anchor text "our services" does far less work than one that links with "roof replacement cost in [region]." Descriptive, specific anchor text is a small habit that compounds across 15 or 20 pages into a real signal.
Note: the deep mechanics of how backlinks, keyword targeting, and technical SEO turn this structure into rankings live outside content strategy itself; content builds the material, the ranking work puts it to use.
One more failure mode worth naming: a silo that links internally but never gets a single external link pointing back at it from anywhere else on the web still caps out below where it could rank. Internal structure earns you the right to compete; outside signals (reviews, citations, the occasional press mention or partner link) are what push a well-built silo the rest of the way on competitive terms. That outside layer is SEO's job, not content's, but it's worth knowing the silo alone isn't the finish line.
Who Should Write It, and What It Costs
Cheap, generic content is the most common reason a contractor's first attempt at a blog failed to earn a single lead. A $25 article written by a freelancer with no trade knowledge reads like it, and it reads that way to an AI model checking whether your content demonstrates real expertise just as clearly as it reads that way to a homeowner. Trade accuracy isn't a nice-to-have here, it's the entire value proposition of the content: a foreman-accurate description of a flat roof recover versus a tear-off, or a correct explanation of a SEER rating, is what separates a page that earns trust from a page that gets skimmed and abandoned.
That doesn't mean every article has to be written by a working tradesman. It means whoever writes it has to actually understand the trade well enough that a foreman reading it wouldn't wince, and ideally the drafts get a review pass from someone who does the work. Some shops handle this with an in-house writer paired with owner review, others use an agency that specializes in trade-accurate content, but generic marketing copywriters with no vertical knowledge are the single most common source of content that never earns a ranking or a lead.
Watch for the tells of a mill-written page before you pay for it: vague cost ranges with no explanation of what moves the price, generic safety tips that could apply to any trade, and zero mention of the permit, code, or inspection realities specific to your region. A page about roof replacement that never mentions wind-uplift ratings, ice-dam risk, or hurricane-code shingle requirements (depending on where the reader lives) was not written by anyone who understands roofing, it was written by someone who searched roofing.
Cost varies with who's doing the work and how deep the research goes, but a useful mental model is to price per silo, not per post: budgeting for a full 12-to-25-page cluster build over several months, with real trade research behind cost figures and process descriptions, costs meaningfully more than a stack of generic 500-word posts, and it's the version that actually moves rankings and gets cited. Anyone quoting a flat per-article rate with no mention of trade research or internal linking strategy is pricing the wrong thing.