The ratio does not set the gain reduction. The overshoot does.

Above the threshold a compressor maps the amount the signal exceeds it — the overshoot — down by the ratio, so GR = overshoot × (1 − 1/R). The ratio’s whole contribution is that second term, a number capped at 1. The overshoot varies by twenty decibels across a programme; the ratio’s term varies by less than one.

The engine is free and needs no account. At 2 dB over the threshold even an infinite ratio removes 2 dB, because there is nothing else up there to remove. At 10 dB over, 4:1 already gives 7.5; going all the way to 10:1 buys 1.5 dB and the entire remaining dial is worth 2.5. Dropping the threshold 3 dB buys 2.25 dB and buys the same again for every 3 dB after it, so after 3.33 dB the mild ratio has passed what an infinite one could have done. One control saturates and the other is linear, and the one that saturates is the one everybody turns. Then the stages multiply: three panels reading 2:1 are 8:1. And a stage can do nothing at all — a bus compressor at −18 dBFS behind a tracking compressor that lands the peak at exactly −18 never engages, at the peak or the average, and reading the panels one at a time will never show it. Every lane ships a worked example that costs nothing. Writing the report needs credits.