Are more recent statistics more important or...

Is data from a longer period of time more important?

For example I am comparing 7 day an 30 day data on my tracker and while the 30 day data says a particular ad is negative roi the 7 day stats say it is a positive one.

What would you do in this case?

You won't get positive ROI everyday. Ideally what you want is to have multiple ads making you money, so if one thing sucks, then the other thing evens it out.

But to answer your question. Here is my thinking process for situations like these

1. How big is the negative ROI? Do I have any ideas to improve it. If the loss is small or I have ideas I will continue on:
   
2. Did I change anything in the past 7 days? If I did then I should have note of what was changed and be able to compare it.
3. Did anything change with the traffic source? You have tracker. Make placement reports on a weekly basis and compare percent change using excel.
4. Is the advertiser/network shaving less for whatever reason? If you have good relationship with AM you can ask them.

Don't try to overoptimize or optimize too early else you risk losing money big time.

If nothing makes sense then it's probably good to just drop it. Generally good campaigns show themselves pretty early on in the optimization process. Sometimes its good to drop the campaign and start something fresh.

Also, you should have other campaigns testing at the same time in times like these. When you have 1 campaign running at a time you risk getting attached to the idea of making it work. It can make you lose alot of time. Time is more valuable than the money.

If you don't have budget to test multiple at once, then you're probably in the wrong traffic source or GEO and should get something cheaper.

Remove not converting site ids.