"soap will kill them"

You'd ... lie to them? Good idea: "soap will kill them" would be more instrumentally useful than some actually true but more complicated explanation, and once they figure out you weren't truthful (at which point they'd probably have figured out other ways to defend against microbes besides the hygiene hypothesis) you would've taught them another implicit lesson: Don't trust the authorities!

So, two for the price of one, with the latter unlocking after the first lost its usefulness. Smart!

OK, but soap does not work by killing germs: it makes it impossible for them to continue to cling to a surface, so they get washed away.

I wouldn't mention soap. Depending on the particular period of antiquity, it's possible no one would know what you're talking about, apparently: http://en.wikipedia.org/wiki/Soap#Early_history And I have to wonder whether that soap would serve as a disinfectant. Further concerns: they don't have experiments, randomization, or statistics - why should they believe that soap works even if they use it?

I hadn't thought about the voodoo risk!

If I wanted to get more specific, I probably wouldn't list ways to kill germs — I'd draw a connection to the decomposition of animals instead. Then whoever's listening could probably figure out the connection to existing food preservation methods for themselves, and ask themselves whether there were other methods.

You could throw that in too, sure. (I considered listing ways to kill germs, like boiling, pickling, salting, drying, and applying alcohol or alkalis, but figured my run-on sentence was run-onish enough already. Having thought about it more I might try a more concise nudge about the tiny creatures digesting carcasses.)