Methodology

This page explains where SaunaCosts's data comes from, how often it is checked for changes, what is and is not tracked, and how the figures on the statistics page are calculated.

Data sources

Data sources: manufacturer specification listings and EIA electricity rates, checked automatically. Where a certification programme publishes a measured value, that record is preferred to a marketing page, because it is the figure the maker had to stand behind under test. Prices come from retailer listings, recorded with the seller named and the date seen, which is why a price can be blank here while the model is plainly for sale somewhere: a listing this site has not read is not a price it will print. Electricity rates are the residential series published by the US Energy Information Administration, used for the running-cost figures and stated with their period on every page that uses them.

Refresh cadence

This dataset is checked for changes on a weekly cycle. When a source listing changes, the next scheduled refresh picks it up and the site rebuilds automatically.

What is included and excluded

Every model page draws on the same fields tracked across the full dataset:

• Capacity (persons)
• Heater power (kW)
• Warranty (yr)
• Price
• Exterior dimensions
• Power requirements
• Wood
• Placement
• Structure warranty

Not every source publishes every field for every model. When a value is missing from the source, it renders blank on this site rather than being estimated or filled in. A field that does not apply to a model at all is left out of its specification table rather than shown blank, so a blank always means "we looked and the source does not say" and never "this does not apply here".

Derivations

Catalog prices are shown as whole dollars, for example $1,235, because at a glance a price column is for comparing and the cents are noise. Amounts a calculator works out for you, such as a cost per session, keep their cents.

The distribution tables on the statistics page group models using thresholds computed from each spec's own 25th, 50th, and 75th percentile values in the current dataset snapshot, not from fixed thresholds chosen in advance. Where two of those thresholds land on the same value, because many models share it, they collapse into one band rather than producing an empty row, so a spec can show three bands instead of four. Each band is open at the bottom and closed at the top: a model exactly on a threshold belongs to the band below it. The filter checkboxes on the comparison table use fixed, hand-chosen ranges instead, because a filter has to keep the same meaning between refreshes; the two groupings will not always match. Medians and maximums are computed directly from that same snapshot at build time.

Where a cost by state figure appears, it is calculated by multiplying a fixed consumption amount by that state's published residential electricity rate.

Affiliate disclosure

Some links on this site are affiliate links. Read the full affiliate disclosure page for details.