Do test tons of variables. Just don't try them all at once. The possible variables for AB/testing are endless. Just make sure that you only have one variable per test. Otherwise you will have a very hard time determining if one test subject was more successful than another.
Do start with simple tests. There is nothing wrong with starting with a small test and working your way up. A simple test could be testing a CTA button size or color. These simple tests may even have significant results. However, don't be afraid to move to more advanced tests later as you gain experience with A/B testing.
Do have a test running all the time. If you have adequate bandwidth, you can even run two tests simultaneously, such as a layout on one page and a CTA on another page. Running two tests simultaneously will help you to optimize your builder or remodeler marketing more quickly.
Don't limit your testing to one marketing channel. While you have a test running on one marketing channel, such as a blog, you can run a another test on a different marketing channel, such as your building company's Facebook page. Make sure that you use the appropriate metrics for each channel.
Do analyze whether statistics are significant or not instead of simply assuming that they are. It is easy to assume that all statistics will be significant. This isn't always the case, so make sure that you analyze before applying the results to your marketing strategies.
Don't give up if they are not significant. Not all of your test results will be significant. This doesn't mean that you did the tests incorrectly or that there is necessarily anything wrong with your marketing techniques.
Do make changes based on your test results. It isn't enough to run the tests. Now you have to take your new found knowledge and apply it to your marketing. Again, if your results weren't significant, don't give up! Simple move on to testing a new variable.