Why we publish our methodology - and how to tell us a score is wrong

Every comparison site has the same structural problem: it makes money from purchases, and it also decides what you see first. Publishing the methodology is the only honest answer to that, because it is the only one a reader can check.

What we publish

• The complete factor list for every scoring profile, with the weight each factor carries.
• The direction of each factor - whether more or less of it scores better.
• The per-factor breakdown on every product page, not just the final number.
• A data completeness marker, so a score built on a thin panel is visibly thin.
• The source values themselves - the nutrition panel and ingredients we read the score from.

That is enough for a determined reader to recompute a score by hand. It has happened, and one of those recomputations found a real bug.

How we earn, and what it does not buy

Filtizy carries affiliate links to retailers. When you buy through one, we may earn a commission at no extra cost to you.

What that does not do is move a product up a list. Scores are computed from product data by a scorer that has no access to commercial terms, and ranking is by score within a category. No brand can pay for a score, a position, or a badge. If we ever run sponsored placement, it will be labelled as such on the unit itself - not disclosed once in a footer.

The test we hold ourselves to: if a reader recomputed a score from our published factors and got a different answer, that is a bug we have to fix - not a proprietary adjustment we get to keep.

What we cannot verify, and say so

Honesty about limits is part of the method.

• We read declared label data. We do not run laboratory tests, so a label that misstates its own product will be scored as it is printed.
• Prices and offers change constantly. A price on a page is the last price we saw at that retailer, not a quote.
• Reformulations reach shelves before they reach product databases. Dates on our listings tell you how fresh the data is.
• Variant matching is hard. A flavour or pack size occasionally lands on the wrong listing, and that is the error readers catch most often.

Where the data comes from

Three sources, in descending order of how much we trust them.

Anything we could not source, we leave blank rather than estimate. A blank is visible on the page and drags the completeness marker down, which is the honest outcome - an invented value would silently improve a score.

How to challenge a score

Use the report link on the product page, or write to us from the contact page. The report that gets fixed fastest contains four things: the product URL, the specific value you believe is wrong, the value you believe is right, and where you read it - ideally a photo of the back of the pack with the panel legible.

If the pack is right and the score still looks wrong, that is a more interesting message and we want it even more. It usually means a factor is behaving badly at the edges of a category, which is exactly the kind of thing a weight table is meant to let us fix without a code change.

What we ask of you

Treat the score as a filter, not a verdict. Read the breakdown when two products are close. Check the pack when it matters. And when something is wrong, tell us - a catalogue is only as good as the people reading it closely enough to notice.