Thaw
Does it defrost to the state the instructions promise, in the time they claim?
Method
How we test
Published in full, before the results, so you can decide whether our verdicts are worth anything.
Controls
- Every dough is bought at retail price, by us, as an ordinary customer. No press samples.
- Boxes are opened and photographed on arrival, before anything is defrosted, and the packaging condition is recorded.
- Doughs are brought to the same temperature by the same route: overnight in the fridge, then to room temperature for the time the seller specifies.
- The same bench, the same semolina for dusting, the same hands stretching.
- The same margherita on every base: identical tomato, identical mozzarella, weighed, so the topping never flatters one dough over another.
- Two bakes each — one in a domestic oven at its maximum, one in a pizza oven at 450°C — because most people own the first and want to know if the dough survives it.
- Every ball in the pack gets used. Consistency across a box is part of the product.
What we score
Stretch
Opens to 12 inches without tearing, snapping back, or leaving a thick collar.
Oven spring
How much rim it throws in the first ninety seconds.
Crumb
Open and irregular, or tight and bready.
Char
Whether it leopards at 450°C, and whether it colours at all at 250°C.
Taste
Salt, ferment, wheat. Blind, against a control dough we make ourselves.
Consistency
Ball to ball across the box. One good pizza is luck.
Condition on arrival
Still frozen? Still sealed? Still the right shape?
Cost per pizza
Delivered, not headline. The number the seller does not print.
Other people’s ratings and reviews
Until a product has been through our bench, the only evidence we have is other people’s. Here is exactly how we handle it, and what we refuse to do with it.
Customer ratings. Where a seller publishes a star rating we record it per source, with the count and the month we checked, and we never blend two sources into one number. A 5.0 from nine people and a 4.7 from seven hundred are not the same fact, and averaging them destroys the only part that matters. Counts under about twenty are marked as too thin to judge. Where a brand publishes no product rating we record its company-wide Trustpilot or Reviews.io score and label it plainly — that score measures delivery and customer service, not whether the food is any good.
Press verdicts. Separately, we look for reviews by named publications that physically tested the product, and we record the verdict as published, with the month. We do not convert them to a common scale, because they are not on one: most titles publish no overall score for pizza ovens at all, giving section scores or a badge instead, and Which? keeps its scores behind a membership. We paraphrase rather than reproduce, and we link out so you can read the original.
Who gets paid. Every publication we have cited so far earns a commission if you buy through its links. That does not make a review wrong, and we are in the same position ourselves — our own outbound links are tagged so we can see what readers click, and we say so on For brands and in our privacy notice. It is why we mark the funding model on every line rather than leaving you to guess, and why we give extra weight to a publication that states it buys the products it tests instead of accepting them free.
What we will not do. We do not use anyone else’s ratings to order a table, we publish no aggregated score of our own, and we do not mark up any of these figures as review data for search engines. They sit on the page as evidence for you to weigh, not as a ranking pretending to be ours.
Counting additives
Every table can be sorted by the number of declared additives. This is the rule that number follows, published so you can argue with it.
What we count. Each distinct named substance in the declaration that performs a technological function: preservatives, raising agents, emulsifiers, thickeners and stabilisers, antioxidants and flour treatment agents, added enzymes and improvers, humectants, and alcohol used as a preservative. We also count added sugars — sugar, dextrose, glucose syrup — and deactivated yeast, because both are added to do a job the ferment would otherwise do.
What we do not count. Flour, water, salt, live yeast, oil and the food ingredients of the recipe. Crucially, we do not count calcium carbonate, iron, niacin and thiamin. UK law requires those in most wheat flour, so a brand that lists them is following the law, and counting them would penalise the brands that spell out their flour spec while rewarding the ones that stay vague. Where a single improver is declared with its components in brackets — an enzyme blend, say — we count the improver once, not once per component.
What the number is not. It is not a health score and it is not our verdict. Ascorbic acid is vitamin C. Calcium phosphates are a raising agent found in ordinary baking powder. A short declaration is not automatically a better pizza, and we will not know which dough is better until we have baked them. What the number does tell you is how much has been added to make the product survive a shelf, and how far it sits from flour, water, salt and yeast.
Where it misleads. Gluten-free dough is the clear case. Hydroxypropyl methylcellulose and the raising agents in a gluten-free ball are doing the structural work gluten does in a wheat dough — take them out and there is no dough, only a crumbly disc. A gluten-free product with fewer additives is not a cleaner product, it is a differently formulated one. We publish the count on those doughs for completeness and would not rank on it.
Counts are only ever compared inside a category. A dry mix has no water, a chilled dough carries preservative because it is not frozen, and a ready base has already been through an oven. Comparing across those formats measures the format, not the recipe.
Scores are not published until a dough has been baked twice. Until then its record carries the specification and nothing else. An empty verdict column is honest; a filled-in one that nobody earned is not.