The 95% effectiveness value is based on the risk ratio: https://sphweb.bumc.bu.edu/otlt/mph-modules/ep/ep713_associa...

From this comment, which summarized the statistics of the phase 3 trial: https://news.ycombinator.com/item?id=25136588

- The percentage of people in control that had confirmed COVID-19 cases was ~0.79%

- The percentage of people in treatment that had confirmed COVID-19 cases was ~0.04%

- This represents a risk ratio of 0.04/0.79 = ~0.05.

A risk ratio of 1.0 means that there was no reduction in risk, i.e. the treatment had no effect. A risk ratio less than 1.0 means that the treatment reduced the risk (or was correlated with this), and above 1.0 means that the treatment increased the risk (or was correlated with this).

Since the risk ratio of treatment was 0.05, there was a 0.95 reduction in risk, which is where the 95% number comes from.