A boiler page deserves to go live when the information unique to that page outweighs the template around it. In practice that means writing down, for each page type, which fields are unique, which are required, and the minimum score a row needs. This guide gives the field lists and pass marks we start from, and a scoring method you can run in a spreadsheet this afternoon.
It is about websites and data only. We're a marketing service, not a contractor, and nothing below is technical or installation guidance.
Unique fields versus template chrome
Every generated page has two layers: fields that change with the row, and everything else. Only the first layer can justify the page.
Template chrome includes the header, navigation, your phone number, your warranty policy, shipping terms, the "why choose us" block and the footer. It appears on every page and is worth zero points.
Unique fields are facts specific to this row: this part's compatibility list, this model's output range, this code's meaning on this brand, this city's gas utility. They are what Google's spam policies are really asking about when they talk about pages that add value.
A useful rule of thumb: if you printed the page and blacked out every line that also appears on its nearest sibling, at least half the main content should still be readable.
Field lists and pass marks by page type
The tables below are starting points. Adjust the weights to match what your buyers actually need.
Part pages
| Field | Points | Required? |
|---|---|---|
| Part number and manufacturer | 1 | Yes |
| Compatible models (verified) | 3 | Yes |
| Real photo of the part | 2 | No |
| Supersession or replacement history | 2 | No |
| Dimensions or specifications | 1 | No |
| Common fault this part fixes | 2 | No |
| Stock status or lead time | 1 | No |
Maximum 12; pass mark 7, with compatibility required. A part page without a verified compatibility list fails, however many other fields are filled.
Model pages
| Field | Points | Required? |
|---|---|---|
| Model name, brand, type (combi, system, heat-only) | 1 | Yes |
| Output range | 2 | Yes |
| Efficiency rating, sourced | 2 | No |
| Venting options and limits | 2 | No |
| Dimensions and clearances | 1 | No |
| Years in production | 1 | No |
| Common faults seen by your technicians | 3 | No |
| Parts you stock for it | 2 | No |
Maximum 14; pass mark 8. The technician fault notes carry the most points because nobody else can publish them.
Fault-code pages
| Field | Points | Required? |
|---|---|---|
| Code, brand and model family | 1 | Yes |
| Plain-language meaning, sourced | 3 | Yes |
| Likely causes | 2 | No |
| Safe homeowner checks | 2 | No |
| When to call a licensed technician | 2 | Yes |
| Related parts | 1 | No |
Maximum 11; pass mark 7. The "when to call" field is required for safety, not for search.
City hubs
| Field | Points | Required? |
|---|---|---|
| Jobs completed there in 12–24 months | 3 | Yes |
| Gas utility and any utility programs, sourced | 2 | No |
| Housing notes relevant to boilers | 2 | No |
| Real response time from your base | 2 | Yes |
| Permit or safety authority, linked | 1 | No |
| Local work examples, written up | 3 | No |
Maximum 13; pass mark 9. If you can't fill "jobs completed", the city shouldn't have its own page. See multi-city hubs for how we gather this.
Sizing pages
Sizing pages fail more often than any other type, so we set the bar highest. The honest answer to "what size boiler for a 2,000-square-foot house in Regina" depends on insulation, windows, air sealing and the emitters in the home, which is why a heat-loss calculation by a qualified professional exists. A sizing page can still help by explaining that process, showing ranges with sources and saying clearly when a calculation is needed. We usually publish a few strong sizing guides rather than a generated set.
How to score your own dataset
You can run this in any spreadsheet in about an hour for a few thousand rows.
- Add one column per field from the relevant table above.
- Mark each cell 1 if filled and verified, 0 if empty or unverified.
- Multiply by the weights and sum to a score column.
- Add a required-fields check: a column that is TRUE only if every required field is filled.
- Filter: rows with a passing score and TRUE in the required check are ready to publish.
- Sort the rest by score. The rows just below the line are your cheapest wins.
A worked example: a parts seller holds 2,400 part rows. After scoring, 900 pass, 600 score 5 or 6 and lack only a photo or supersession note, and 900 have no compatibility data at all. The sensible plan is to publish 900, spend a few weeks filling the 600, and leave the 900 without compatibility out entirely until a technician or supplier can verify them.
Mistakes that inflate scores
Scores are only as honest as the data behind them. Watch for these.
- Counting copied text as a field. A manufacturer's description pasted in is not your data and earns nothing.
- Unverified compatibility. A supplier's "fits most models" is not a compatibility list.
- Generated filler. A sentence written by a tool to describe a field that's actually empty.
- Duplicated technician notes. The same fault note pasted across 40 models scores once, not 40 times.
- Placeholder photos. A generic boiler image is template chrome, not a real photo.
After the first batch: tuning the line
The pass mark is a hypothesis, and the first batch tests it. Two or three months after launch, compare indexed share and impressions for pages just above the line against pages well above it.
- If pages just above the line are indexed at much lower rates, raise the line.
- If they perform almost as well as the strongest, you may be holding back too many rows.
- If an entire type performs poorly at every score, the template or the page type itself is the problem.
This is where per-type sitemaps matter; see indexing and crawl control. The cost side of the same decision is covered in the build economics guide.
Checklist before you set a threshold
- Each page type has a written field list with weights.
- Required fields are marked, including any safety fields.
- A person who knows boilers has checked a sample of 50 rows.
- Every field has a source column.
- Failing rows have a plan: fill, noindex or leave out.
- The line will be reviewed after the first batch.
If you want help building the dataset under these tables, that is our boiler dataset building service. Call, text or email with a sample of 20 rows, and we'll score it with you.
Frequently asked questions
Is there an official minimum word count for pages?
No. Google doesn't publish a word count, and word count is a poor proxy anyway. A part page can be short and useful if it carries the fields a buyer needs; a long page can be empty.
What score should a row need before it goes live?
We usually start at 60% of the available points for that page type, with certain fields marked as required. Adjust the line after the first batch shows which pages Google indexes.
Can I publish rows that fail and improve them later?
You can publish them with noindex so readers can still browse them, or keep them unpublished. Either is better than submitting them for indexing while they're empty.
Who should check the data?
Someone who knows boilers. A senior technician or parts manager spotting errors in a sample of 50 rows is worth more than any automated check.