Branko Rihtman of SEOScientist presented an interesting idea about how to test SEO:
He calls the method “Multi-directional” experiments. Basically here’s how it works:
- Benchmark where you are (e.g. rankings/traffic for a particular keyword)
- Test your theory by changing from State A to State B
- Wait a bit (at least until the changed pages have been crawled)
- Compare results v. benchmark
- Change State B back to State A
- Wait a bit (at least until the changed pages have been crawled)
- Compare results v. #4 & benchmark.
- If the change back to State A is similar to where you started then this might mean that your test is valid.
- Use a professional statistician to help validate your data
- Avoid personal bias
Other ideas:
When you test links, link not only to your website, but also to a website you have no control over. If the other website behaves the same way as yours after the link is crawled, then that is a good sign that the link is working.
3 Response Comments
I suppose that if the change back to State A is not similar – this means there would be a boost regardless, so the link can’t be proven to have helped?
Of course, make sure you revert to state A before publicizing your test, as the publicity/links drawn from that would be likely to then impair your test.
This is a great, simple, approach. Thanks for sharing.
I especially like the link testing ideas and have done some of that in the past. Probably a good time to apply again and see how the algos in 2010 respond.