Why Do Ab Testing
title: Why do A/B testing? description: A/B testing has many uses, helps your business, improves your product line, gives you insights into best index and query configurations.
We speak a lot about relevance - getting the best results. Relevance is nearly everything when it comes to search. That’s why the vast majority of our API methods and settings, as well as many of our Dashboard features, are devoted to giving you direct control over the relevance of the results.
But before you start thinking about relevance, your first concern is to get your search up and running. For this, you need to:
- Push your data
- Create you search UI
- Configure relevance (eg, set up searchable attributes, add custom ranking)
Once the solution is delivered, however, you can and should take a step back and look closer at the results and ask the right questions: Is this what you want to see? Are some items appearing too often, not enough, too far down in the results, or not at all? These are all questions about relevance. To get better results, you can:
- Tweak the engine’s default settings (typo tolerance, language-handling, et..)
- Set up our data and attributes differently
- Add synonyms, query rules, filters, etc.
- Re-order our results with custom ranking and alternative sorting strategies
- And more …
While this kind of work can get quite advanced, it is absolutely necessary if you want a great search experience. Only after working with the search engine over time will your configurations and data reformatting make sense. This is the experimental stage, where tweaking and experimentation are best practices for you and your end-users. This is a worthwhile and rewarding process, and it doesn’t end - when new products and new trends appear, and customers and needs change, and when what was good yesterday is no longer good today.
As soon as you start using A/B Testing, it will become instrumental to this experimental process. It ensures that your experiments are carefully set up, that they make sense, and that the conclusions drawn from them are sound and reliably data-driven. A/B Testing involves actual user feedback. This feature turns a somewhat challenging process into a simpler and more accurate one.