"But Linda, didn't you say you should test checkout first?"
Sounds like something I'd say. ;) I believe you should absolutely test your checkout before home page, category pages etc. (in most cases). I would vouch for testing a bit of everything all at once, but there are limitations to A/B and multivariate testing platforms that prevent us from doing so. You can usually only test one thing at a time.
However, you can test an unlimited number of PPC landing pages at the same time, as they do not require special tracking code. You can create custom URLs that are tied only to your individual paid search ads. These URLs will be inaccessible from your main site.
So they answer is, you can test both checkout and PPC landing pages *first.*
Why test PPC pages first?
1. You're paying for this traffic. Pages that don't convert are sucking money on a constant basis. For competitive keywords (high search volume, high competition, high click through and high cost-per-click) the cost can be astronomical.
2. You can predict purchase intent by referring keyword. (This is not possible in Google Website Optimizers and some other testing platforms). When you can control for what the purchase intent is (e.g. "accounting software for small business" vs "accounting software for non-profit"), you can better craft messaging. Both searches may point to the same product, but the highlighted features, headline, messaging and testimonials can all be laser-targeted to the end-user.
3. You can optimize for the medium. Paid search lands directly on the product or offer page, which means the visitor has not navigated through your site and may not be familiar with your value proposition, offering, etc. You can apply the learnings you've gleaned from web analytics about paid search vs. rest of site traffic to these pages (use your Advanced Segments in Google Analytics or other platform).
4. When you optimize for the most popular product pages, you'll drive more traffic to your checkout (and your checkout test, which may speed test completion there as well).
5. You can run unlimited tests concurrently (I've said it once before but it bears repeating). Plus, your learnings may be applicable to the rest of your site (e.g. headline A converts better than headline B across a number of tested keywords). This is efficient!
Where to start? Start with the keyword(s) / AdGroup(s) that cost you the most money for the least reward, of course! It's important you look for the combination of high traffic and low conversion. (Hint: use weighted sort if you're using Google Analytics, or similar feature in the analytics package of your choice).
Interested in improving your ROI from paid search? Contact email@example.com to learn how a pay-per-click audit can help improve your business results.