I wonder if the 95% effectiveness was actually true: against the strains prevalent at the time they ran the trial. But effectiveness against later strains was diminished.