I like to see 1000 impressions before making a decision. Make sure you're not jumping too quick.
If you've got a good ad, change small things only as you test. i.e 'fast and easy' will perform differently than 'easy and fast'. Even small changes like that can make a difference.
Don't test randomly. Know what changes you're making to the next ad you're testing, and why.
The question I have is:
Are you testing CTR?
OR testing conversion rate/profit by ad?
A high CTR ad is often not as good as a low CTR higher converting ad.