How We Rank, and What We Refuse to Claim

1. What we do

Peel & Flame ranks pizza equipment on four things, in this order:

Published specifications. Dimensions, mass, material, thickness, stated maximum temperature, and the maker's own usage guidance. Every specification on this site renders with a link to the page it was read from. Where a maker publishes nothing, the row says exactly that rather than carrying a number we supplied.

Thermal arithmetic from material constants. Preheat behavior, stored energy, effusivity and heat capacity, computed from the constants table in section 2 and the model in section 3. Figures that are ours rather than the maker's are labeled as derived wherever they appear.

Manufacturer manuals and guidance. Read, not paraphrased from somebody else's summary of them.

Live retailer data. Price and availability, checked on the day you are reading, through the price layer described in section 6.

2. What we do not do

We do not test. We have not baked on any of these products. There is no test kitchen, no lab, no instrumented protocol and no owned equipment. Every competing guide in this category claims otherwise; several of them describe six-week protocols and sixty pizzas while publishing no photographs, no per-unit data, no temperature logs and no named tester. We would rather state the absence than assert something unfalsifiable.

We accept no free products. None have been sent to us and none would be accepted.

We take no payment for placement. Not for a ranking, not for inclusion, not for a mention.

We do not publish a rating in structured data. The scores on this site are editorial and visible, and they never appear as Review or aggregate-rating markup - because marking our own opinion up as a product rating would be a machine-readable testing claim we cannot support.

We do not write prices into pages. A price typed into a page decays into a false statement the moment it changes. See section 6.

3. The material constants table

One table, one source per figure, imported by every page that needs it - so no two pages on this site can quote different numbers for the same material.

Derived quantities

These are ours, computed from the table above. Volumetric heat capacity is density times specific heat. Effusivity is the square root of conductivity times volumetric heat capacity. Diffusivity is conductivity divided by volumetric heat capacity.

The ratios that follow, steel against cordierite: conductivity 33 times using the midpoint of cordierite's published range, and 29 to 38 times across that range. Volumetric heat capacity 2.1 times. Effusivity 8.5 times. Diffusivity 15.6 times.

Cast iron sits within 5 percent of carbon steel on effusivity, which is why a cast iron pan behaves so much like a baking steel.

4. The thermal model, and its one weak assumption

Preheat behavior on this site comes from a lumped-capacitance model: the whole plate is treated as a single temperature rising toward the oven's, limited by how fast heat crosses its surface rather than by how fast it travels inside.

Step 1 - check that the model applies

Lumped capacitance is valid where the Biot number - h times the characteristic length (volume over surface area) divided by thermal conductivity - is below about 0.1. Every baking surface on this site is a thin plate with large faces, and every one comes in at or under that threshold. A quarter-inch carbon steel plate lands at about 0.001; the worst case, a thick cordierite stone, lands near 0.09. So conductivity - the property steel wins on overwhelmingly - is not the limiting factor for any of them.

Step 2 - the time constant

tau = m x cp / (h x A)

where m is mass in kilograms, cp is specific heat capacity in J/kg.K, h is the effective surface heat transfer coefficient in W/m2.K, and A is the total exposed area in m2. After one tau the surface has closed 63 percent of the gap to oven temperature, after 3 tau 95 percent, after 4.6 tau 99 percent.

Step 3 - the assumption you should challenge first

h = 20 W/m2.K. This is the only input in the whole model that is not a published constant. It is the sum of a linearized radiative component - for a 0.9-emissivity surface warming toward a 500 F cabinet, that works out near 21 W/m2.K - and a few W/m2.K of still-air natural convection. Real domestic ovens plausibly sit anywhere in a 20 to 30 W/m2.K band, and at 30 every preheat figure on this site shortens by about a third.

We use the conservative end deliberately, so our published times are if anything long rather than short.

A fully worked example you can check

Baking Steel Original, 16 by 14 inches, quarter inch thick, published mass 15 lb.

1. Mass: 15 lb x 0.45359 = 6.80 kg.
2. Faces: 0.4064 m x 0.3556 m = 0.1445 m2 each, so 0.2890 m2 for two. Edges: perimeter 1.524 m x 0.00635 m = 0.0097 m2. Total exposed area A = 0.2987 m2.
3. Specific heat capacity of carbon steel, from the table: 470 J/kg.K.
4. tau = 6.80 x 470 / (20 x 0.2987) = 3,196 / 5.974 = 535 s = 8.9 minutes.
5. 95 percent of oven temperature: 3 x 8.9 = 27 minutes. 99 percent: 4.6 x 8.9 = 41 minutes.
6. Energy to go from 68 F to 500 F (a rise of 240 K): 6.80 x 470 x 240 = 767,000 J = 0.21 kWh.

The cross-check. Baking Steel publish their own preheat guidance for this plate: 45 to 60 minutes at 500 to 550 F. Our 41-minute figure to 99 percent was arrived at from their published mass and a textbook specific heat capacity, with no knowledge of their recommendation. Two independent routes landing on the same answer is the strongest validation available to a publisher that owns none of this equipment.

A finding that came out of the method

Published density for a dense cordierite body is 2,000 to 2,200 kg/m3. Applied to the two stones on this site whose retailers publish both dimensions and a weight, that density overstates the mass by about 20 percent in both cases - and solving backwards gives an effective density near 1,700 kg/m3 for both, agreeing with each other to within one percent.

That is porosity, and it means a real bakeware stone holds about a fifth less heat than its dimensions suggest. Where a published weight exists we use it; where one does not, the derived mass is labeled as an upper bound on the page.

5. How ranking works

Weighted, in this order, and stated so you can disagree with the weighting:

1. Fitness for the actual job - does the thing solve the problem the page is about, for a domestic kitchen. A commercial dough box that will not fit a fridge shelf loses here regardless of how good it is.
2. The physics, where physics applies - effusivity, mass, preheat behavior, computed as above.
3. Published specification quality - a maker who publishes a weight, a thickness, a temperature limit and a preheat procedure has measured their own product. One who publishes a photograph and a slogan has not. This is a genuine tiebreaker on this site.
4. Practical handling - weight, storage, cleaning, and whether the thing will actually get used rather than admired.
5. Durability and failure modes - what breaks it, how likely that is, and whether it is recoverable.

What would change a ranking: a maker publishing specifications they previously withheld; a product being discontinued or superseded; a correction to one of our own figures; or a change in the constants table. Price is not a ranking input, because we do not publish prices - though value shows up indirectly, since a product that does the same job with less mass and fewer requirements tends to score better on handling.

6. How prices work

Prices render only from a live data layer, never from copy. That layer has a 48-hour freshness gate: any figure older than that disappears and the button falls back to "Check price on Amazon". It never shows a stale number, and there is no code path by which it can show a remembered one.

At the time of writing, the site holds no live price data at all - the retailer API credentials are not configured for this property - so every buy button reads "Check price" and no price appears anywhere on the site. That is the system working as intended rather than a gap: the alternative would be numbers we could not stand behind.

The same applies to product photography. Where we have no verified image of a specific model, we show a licensed photograph of the product type and say so in the caption and the alt text. Nothing is passed off as a picture of a particular unit.

7. Energy and cost figures

Where we publish a running cost, it is built from three inputs, all stated on the page:

• Element rating: 2.4 kW, a typical domestic electric bake element. Yours is on the appliance plate.
• Duty cycle: continuous for the first 15 minutes, then roughly 30 percent to hold. An assumption, and one you can check with a clamp meter.
• Electricity price: 18.34 cents per kilowatt-hour, the US average residential retail price, June 2026, from the US Energy Information Administration, Electric Power Monthly, Table 5.3. Your own rate is on your bill and is the number most worth substituting.

8. Sourcing standards

• Every physical or numeric claim carries a source, or is our own arithmetic shown inline.
• Third-party findings are attributed in the sentence, with the publisher named.
• Ranges stay ranges. Where the underlying data spans a range, we publish the range rather than a midpoint dressed up as a measurement.
• Retailer data is labeled as retailer data, with the retailer named. It is never presented as our own assessment.
• We do not source a number to another affiliate site's unsubstantiated testing claim. If an unverifiable review page is the only source for a figure, the figure does not go on this site.

9. Corrections

We will get things wrong. When we do:

Report it to info@peelandflame.com. We respond within five business days. Where you are right, the page is corrected and carries a dated correction note - we do not silently edit a page and pretend it always said the new thing. Where we think the original was right, we will tell you why.

10. Update cadence

Every page displays a real published and updated date, and the XML sitemap uses those dates rather than the build timestamp - so a redeploy does not claim that every page changed.

11. Who writes this

Scooter M., an enthusiast. Not a chef, not a food scientist, not a credentialed anything - and the editorial policy lists in public every claim we have banned ourselves from making, including all of those.