How We Pick
The whole method, written out. No competitor in this category publishes one, which is exactly why this page exists.
What we do not do — stated first, on purpose
- We do not test the products. We do not own them and we have never used them. Any page implying otherwise would be lying to you.
- We accept no free units and no sponsored placements. Nothing on this site is purchasable.
- We hold no credential in this field and claim none. No ornithologist, no wildlife biologist, no expert panel, no “reviewed by”.
- We publish no rating of our own. There is no score on any product page and no rating markup in the structured data, because we have collected no ratings to aggregate.
That is a smaller claim than the field makes and it is one you can verify. Every number below links to where it came from.
How a pick gets chosen
In this order:
- 1. Published specification fit for the winter use case. Basin width and depth, mount type, cake capacity, seed capacity, wall thickness — whichever applies. Read off the manufacturer’s own listing and cited on the product card.
- 2. Rated wattage, and whether it exists at all. For anything electrical this is the deciding criterion, because it is what you pay for every day of the winter. Currently 9 products in our registry carry a published, cited wattage, and 5 state none at all.
- 3. Thermostat behavior and temperature rating, where the maker publishes it. Erva is the only maker in our registry that prints its set points, and we cite them.
- 4. Computed running cost, at the rate below, shown with the working.
- 5. Build against freeze-thaw. Cast aluminum and metalwork over molded plastic; inch-thick pine over half-inch cedar. Where the listing states a material, we use it; where it does not, we do not assume.
- 6. Cord length and outdoor power practicality. The constraint most buyers discover after the box arrives.
- 7. Current price and availability from the live data layer.
Every pick also carries a skip it if line, and every roundup names at least one product we would not buy. A ranked list where nothing has a downside is an advertisement.
How to check our work: the running-cost formula
This is the site’s signature asset and it is deliberately reproducible. The entire calculation is one line:
watts ÷ 1000 × hours × your rate per kWh = dollars
We use $0.1834/kWh, which is US average residential retail price of electricity for June 2026, published by the US Energy Information Administration, Electric Power Monthly, Table 5.3 (June 2026): 18.34 cents per kWh, US average residential. Retrieved 2026-08-31. Substitute your own rate from your electricity bill and every figure on this site recomputes.
The caveat we will not remove
Every figure we publish assumes the element draws its rated wattage continuously. Nothing does — every unit here is thermostatic and cycles off once the water is clear. We could multiply by a duty-cycle factor and print a lower, friendlier number, and we are not going to, because we have not measured a duty cycle on any of these units. A plausible-looking figure with no measurement behind it is precisely the thing this site exists to avoid. The rated wattage is what the manufacturer stands behind, so that is what we compute from, and we tell you it is a worst case. The full table is here.
Where the numbers come from
- Wattage, capacity, dimensions, materials, cord length: the manufacturer’s own listing, linked on every product card. Never an estimate, never a remembered figure.
- Prices: a live data layer refreshed daily. Any price older than 48 hours disappears from the site automatically and the page falls back to “Check price”. There is no path by which a stale number can be rendered.
- Electricity rate: EIA Electric Power Monthly, cited and dated above.
- Bird behavior, physiology and species: named authorities — the Cornell Lab of Ornithology, Project FeederWatch, Audubon, the US Fish & Wildlife Service — linked in the sources section of every page that makes such a claim.
Where a claim cannot be sourced, it does not get published. Not softened, not hedged — omitted.
When a pick changes
- The published specifications change.
- The price changes enough to alter the recommendation.
- A product is discontinued or goes into a sustained stock-out.
- A better-documented product appears in the category.
- A reader shows us we got something wrong.
Never because a commission rate changed. Where the cheaper or better option for the buyer pays us less, that is still the option we link. The editorial policy covers corrections and update cadence.
Seasonal refresh
This is a seasonal category, so the calendar is part of the method. Every roundup is reviewed before 1 October, ahead of the season’s demand. The running-cost table is revalidated whenever the EIA publishes new figures. The “Updated” date on every page is a real content date, never a build timestamp — a date that moves every time the site redeploys is a date that means nothing.
How we make money
Affiliate commission on outbound product links. The full disclosure sets out which merchants, what we earn, and what it does not buy.