A programmatic page system is three things: a map of the page types you will publish, a template for each type, and a pass mark that decides which data rows become live pages. Boiler companies are well suited to it, because the trade runs on model numbers, part numbers and fault codes that people search for one at a time.
We build the system around your own data. We do not install or repair boilers and we are not a contractor; our work ends at the website, the data and the search structure.
What a page system is, and what it isn't
It is a catalogue, generated from structured data, in which every page has a different job. It isn't a way to multiply one article into hundreds of URLs.
Google's spam policies describe scaled content abuse as producing many pages mainly to manipulate rankings, with little original value, however the pages are made. Templates are not named as the problem; low-value pages are. So the design question we ask for each page type is: what does a reader get on this URL that they could not get on the one next to it?
For a boiler business the strongest answers usually come from:
- Parts: the exact models a part fits, its supersession history, and photos of the real part
- Models: output range, efficiency, venting and clearance data, and the faults technicians see most
- Error codes: what the code means on that brand, likely causes, what is safe for a homeowner to check, and the point at which a licensed technician is needed
Who it's for
The service fits boiler businesses that hold a real library of distinct items.
- Parts sellers and distributors with hundreds or thousands of SKUs
- Installers and service companies that work across many brands and keep notes on recurring faults
- New boiler divisions of larger HVAC firms that need a structured site from day one
It doesn't fit a single-town installer with twelve services. For that business, a handful of strong service pages and a city hub or two will do more; see multi-city boiler hubs.
What's included
Each engagement delivers the same core pieces, scaled to your catalogue.
- Page-type map. Every type you could publish, the number of rows behind each, and our recommendation on which to launch.
- Field inventory. For each type, which fields are unique per page, which are shared, and which are missing.
- Templates. Layout and copy for each type, with written sections that explain and data sections that change per row.
- Pass mark. The minimum populated fields before a page can publish, written down and applied by the build.
- Internal linking rules. Parts link to the models they fit, models link to their common fault codes, and fault codes link back to the parts that fix them.
- First batch. Usually 100 to 300 pages, published and submitted in their own sitemap so we can watch them separately.
The dataset itself is covered by boiler dataset building, and the indexing rules by indexing and crawl control.
Timeline
A typical first launch runs eight to twelve weeks from data handover.
| Weeks | Work |
|---|---|
| 1–2 | Data review, page-type map, pass mark agreed in writing |
| 3–5 | Templates designed and tested against 20 sample rows, including the weakest ones |
| 6–8 | First batch generated, reviewed by a person, published |
| 9–12 | Indexing watched; template fixes; decision on the next batch |
Large catalogues are released in stages after that. Publishing 2,000 pages on one day tells you nothing about which template works.
How we measure results
We judge each page type on its own numbers, not a sitewide total.
- Indexed share: of the pages submitted in that type's sitemap, how many Google chose to index
- Pages with impressions: how many earn any search visibility within 90 days
- Pages with clicks: the smaller set doing real work
- Calls, texts and orders traced to pages of that type
If a type has a low indexed share after three months, that is a signal about the pages, and we treat it as one. The programmatic build economics guide shows how to turn these numbers into a cost per useful page.
Mistakes to avoid
These are the errors we see most in boiler page sets built elsewhere.
- City-times-service grids. "Boiler repair in [town]" multiplied across 80 towns, with the same paragraph on each. This is the pattern Google describes as doorway abuse.
- Publishing empty rows. A model page with a name, a stock photo and nothing else.
- Copying manufacturer text. Pasting the brochure paragraph onto every model page adds nothing a reader can't find on the maker's site.
- No way back. Building pages without a plan to noindex or remove the ones that never earn visits.
- Safety overreach. Error-code pages that coach homeowners through gas-side repairs. Our templates stop at safe checks and send the reader to a licensed technician.
Our own reference build is BoilerPart.ca, with 2,000+ programmatic pages and 180% organic traffic growth; the results page explains what made it work. Plans and costs are on the pricing page.
Frequently asked questions
How many page types should we launch with?
One or two. Launch the type where your data is strongest, usually parts or models, watch how it is indexed for two or three months, and add the next type only when the first is working.
Can you generate pages for boiler brands we don't sell?
Only if you hold accurate data on them and a reason to help those buyers, such as stocking compatible parts or servicing that brand. Pages for brands you know nothing about tend to be thin by definition.
Will the pages look identical?
The layout will be consistent, which helps readers. The content inside it changes with the data, and we test sample rows to make sure two neighbouring pages never read as the same page with one word changed.
What platform do you build on?
We work with whatever you run where it can handle structured content, and recommend a change only when the current platform can't generate pages from a dataset or control indexing.