Can I assume you mean log transform the outcome? This is certainly an appropriate approach if your outcome is not normally distributed. When you do so, the difference-in-difference estimator (the coefficient on the interaction term [between "treatment" and period 2]) is interpreted as a net percent change in the outcome. (See chapter 13 in Jeff Wooldridge's "introductory econometrics" for details and examples).

I hope this helps

Ariel

Thanks Ariel. I did exactly that. I was not sure if it was right. I will check woodridge.

Just keep in mind that the sensitivity of log is very high for values between 0 and 1 and very low above 1.