The % effectiveness is reported as a point estimate eg 70%.
What that means is that if 100 people on the placebo arm got covid, 30 on the active did. So if 100 people would get the disease unvaccinated, 70 of those would have been saved by the vaccine, hence 70% effective.
You're absolutely right that you can't actually give a precise estimate with such relatively small numbers. A confidence interval, calculated statistically is used. For the oxford jab, the confidence interval is 54% - 80%, the estimate of the real efficacy. This means we can be 95% sure the true efficacy is in the range 54-80%. Even a 54% effective vaccine is more than good enough to be worthwhile, so we can be near certain this is sufficiently effective. We can't be certain what the exact figure is.
Does that help?