In this chapter, we learned how to correct user spelling mistakes both by using the terms suggester and the phrase suggester, so now we know what to do in order to avoid empty pages that are a result of misspelling. In addition to that, we improved our users' query experience by improving the query relevance. We started with a simple query; we added multi match queries, phrase queries, boosts, and used query slops. We saw how to filter our garbage results and how to improve the phrase match importance. We used N-grams to avoid misspellings as an alternate method to using Elasticsearch suggesters. We've also discussed how to use faceting to allow our users to narrow down search results and thus simplify the way in which they can find the desired documents or products.

In the next chapter, we will finally get into performance-related topics, starting with discussions about Elasticsearch scaling. Then, we will discuss how to choose the right amount of shards and replicas for our deployment, and how routing can help us in our deployment. We will alter the default shard allocation logic, and we will adjust it to match our needs. Finally, we will see what Elasticsearch gives us when it comes to query execution logic and how we can control that to best match our deployment and indices architecture.