Neither
Taking Max's (great) answer further:
Who on earth measures weight changes in percentages?
The problem with percentages
- In general, percentages are always a proxy variable and mostly little intuitive.
- Everyone has to expand some cognitive effort to do the maths, and quite a few struggle with the percentages maths (some can't do it at all).
- When time is a factor, percentages get ambiguous - if you lost 2% on week one and 2% on week two, did you lose 4%? Was the base rate for the second measure the weight on the first or second week?
- Percentages are relative - if two people lost each 10% in 3 months, it is much more of an achievement for someone who weighted 160kg (lost 16kg) than to someone who weighted 60kb (lost 6kg).
The benefit of percentages
Percentages are great when there is a fixed lower limit (often 0), and fixed upper limit. If data allowance is 2GB per month, most people would prefer to know they have used 80% of it, than 1.6GB. I'm afraid weight neither has an absolute lower limit, nor an upper one.
Another place where percentages are great is when you compare change between figures that has different reference point. So if Apple stock went up 5% and so did Google's - their performance is on par; much easier than being given the start/end stock value (which in itself depends on the amount of stock shares). But I doubt you aim to compare such change.
Summary
The point is that in everyday language, weight changes are measured in Kgs, lbs, or stones. So what is the case for percentages?