Traditionally, search engine optimization (SEO) tactics include trying out different known strategies and hoping for the best, which often makes people think of SEO as magic rather than engineering. I would like to share our story of how our Pinterest engineering team turned this “magic” into deterministic science with the help of data science and a pinch of curiosity.
Since Pinterest published a blog post about how we build the SEO testing tool back in 2015, many big and small tech companies adopted this idea. However, SEO testing is still tricky to set up depends on the product and there are lots of challenges still to be explored.
In this talk, I will outline the basics of SEO experimentation framework with some common use cases and challenges by covering the following topics:
- Setting up page-level A/B test
- How do the search engine bots and users see the A/B test differently
- Dilution, carryover effect and novelty impact on SEO A/B test