The trade counts kill in logs, and the public reads it in percents. They are the same statement, and the conversion is worth knowing cold: $P = 1 - 10^{-\mathrm{LR}}$ — P equals one minus ten to the minus L-R. $P$ is the fraction destroyed, running from 0 to 1 (multiply by 100 to say it as a percentage), and $\mathrm{LR}$ is the log reduction, the bare count of powers of ten. The term $10^{-\mathrm{LR}}$ is the surviving fraction, which is what makes the identity readable: kill plus survivors is one.

The table is short and worth memorising. 1 log is 90%. 2 logs is 99%. 3 logs is 99.9%. 4 logs is 99.99%. For any claim written as a run of nines, the log reduction is the count of the nines.

Now the reason professionals prefer logs. Percentages crowd together at the top of the scale: 99% and 99.99% look almost identical written out, and differ by a factor of a hundred in survivors. Logs keep those apart, because each one is a whole step. And the two forms must never be added — writing "4-log removal plus 99.99% removal" counts a single barrier twice, which is a real and recurring error in treatment reports.

Going the other way, $\mathrm{LR} = -\log_{10}(1 - P)$: take the survivors, take their logarithm, flip the sign. And to find what is actually left, $N = \dfrac{N_0}{10^{\,\mathrm{LR}}}$, where $N_0$ is the count going in and $N$ the count coming out.

Which brings up the one thing a log reduction cannot tell you on its own: it is relative. Twelve logs from a starting load of $10^{3}$ per can and twelve logs from $10^{6}$ leave wildly different numbers of survivors. The incoming load matters as much as the process does, which is why raw-material hygiene is part of a thermal schedule and not a separate conversation.

The famous number in this field is 12D — the classic low-acid canning convention, twelve decimal reductions against the target spore. Read it as exactly what it says: twelve D-values laid end to end, $t = 12D$. Nothing more mysterious than that, and nothing less demanding.