Why We Publish Our Formula (and Most Ratings Sites Do Not)

Look up any school on a big ratings site and you will find a number, a color, and a paragraph about "proprietary factors." What you will not find is the formula. You cannot recompute the rating, so you cannot argue with it, so you cannot really trust it. You can only obey it.

We think that is backwards, so our methodology is public: which files, which fields, which weights, which thresholds. Anyone with a spreadsheet can rebuild any grade we publish. Here is why we chose that, honestly, including the downsides.

What transparency buys families

Arguability. A parent who thinks a grade is unfair can check the math and the source file. Sometimes they will find a data problem, and we fix it and say so. A black box can never be wrong, which is exactly why it should never be trusted.

Context. Because the formula is visible, its limits are visible too. Our grades are built from state assessment results; they cannot see school culture, safety, or fit, and test scores correlate with demographics. Published limits are a feature. Hidden limits are a liability that gets families into wrong schools.

Accountability for us. Our vendor directory has a disclosed featured partner. The strongest protection against even the appearance of influence is that grades are mechanically recomputable from public files by anyone. Nothing about a grade can be nudged without being caught.

What it costs us

A published formula is criticizable, and it will be criticized: a proficiency-and-percentile formula is simple, and simplicity leaves things out. Schools serving special populations get flagged instead of graded, which some will call dodging. And competitors can copy the method wholesale, since it is right there.

We accept all three. Criticism improves the formula in public, flags are more honest than false precision, and a copied methodology still needs the data pipeline, the directory, and the trust, which are the actual product.

The standard we are proposing

Any school rating you rely on should answer three questions in writing: What data? What formula? What changed since last year? If a site will not answer, treat its number as marketing. Ours are answered on one page, and when the formula improves, the change and the reason will be published before the new grades are.