In this sprint we will configure our relevance within Algolia, using only the dashboard to ensure our users are returned relevant results.
Depending on the size of your company, some of these roles may be the same person. This sprint it is important we identify these roles and get in contact with them.
Planning and project oversight
Product vision, planning, prioritizing and management lifecycle
Management of e-commerce product display
If we have indexed our data so we have one index per language we can optimise our relevance by setting removeStopWords, ignorePlurals (this can be done directly or by setting the index) index language and query language.
For Danish, German, Finnish, Dutch, Norwegian, Swedish, we can set decompoundedAttributes and decompoundQuery. If we are not seeing the decompounding we are expecting we can review in the custom dictionaries.
Algolia does support language transcriptions through the customNormalization setting.
For example, in German, Umlaut vowels (ä, ö, ü) are commonly transcribed with ae, oe, and ue if the umlauts are not available on the keyboard or other medium used. In the same manner ß can be transcribed as ss.
Iteration and review
Relevance should never be viewed as ‘finished’ there is always optimisations that can be made. It is recommended to frequently test top queries and if they return less desirable results, to troubleshoot them.
Synonyms tell the engine which words and expressions to consider equal. They come in many different configurations.
Personalisation could be a good fit for you if your users tend to return and want to access similar items, eg books from the same author, dresses from the same brand or documents from the same project.
In order to implement personalization you need to ensure that the userToken is being sent with queries. A good way to check this is within the events hub, it’s important to have validated events. If all events are valid personalisation can be enabled and the strategy can be set. It is recommended to set live within an AB test so we can see how effective it is.
Algolia identifies queries that your users often change and proposes synonyms for them. These can be accepted and rejected based on your understanding of what your users are searching for.
AI Re-Ranking is an Algolia feature that leverages AI to find trends in your users’ behavior. Based on the query and the position of the result they click or convert, it can make improvements to your relevance by boosting results that are rising in popularity. To enable AI reranking, you need to validate your events and then you can start to test your reranked queries in the Re-Ranking Simulator. Once you are happy with the results, you can launch an AB test and then go live for all users.