Why should you do it?

Once you have configured your global relevance strategy and implemented manual and automatic merchandising strategy, it’s time to analyze and optimize the performance. The analytics data that can be accessed through Algolia dashboard can supply you with multiple insights about your strategy. In this section, we are going to concentrate on understanding the analytics data, analyzing the insights and acting on them in order to achieve the organization’s digital merchandising goals.

Low CTR and CVR mean that the percentage of searches that led your users to take a conversion action is too low and needs improvement.

Optimize this metric:

• Check click position - the average position of the clicks performed on the search results. Smaller values are better since they indicate that users tended to click higher up in the results. If the click position is too high, the results order should be changed.
• Use AI re-ranking, AI synonyms and Personalization to optimize CTR and CVR. Make sure to A/B test your new strategy.
• Use merchandising Rules to fine tune and optimize the results and conversion by making precise and (if desired) temporary modifications to your search results.
• Review and optimize your current solution (as in the case of handling no results)

Abnormal and unusual CTR or CVR might hint that the Click and Conversion events are not implemented properly and require additional validation.

Optimize this metric:

• Validate the you send the Click and Conversion events correctly:
• Events Hub section of the dashboard under the “Platform” tab, provides you with a snapshot of searches and events recently sent to Algolia to ensure you get the most out of your data. 
• Debugger section of the dashboard, under the “Platform” tab, displays events as the Insights API receives them.

Note: You might need assistance from the engineering team to validate Click and Conversion. Refer to “Validating events” documentation guide for more details.

A high no clicks rate means high percentage of searches that retrieved results didn’t generate clicks.

Optimize this metric:

You can apply the following tactics to optimize the order of the search results presented to each user:

• AI Re-Ranking: to ensure the most relevant results appear at the top, while the least relevant are pushed to the bottom
• Personalization: to ensure that each user will see the most relevant results for their unique preferences
• Suggested Searches: to suggest popular search queries to your users and guide them to find the most relevant and popular search results. Ensure that your developers team has implemented a query suggestion index.
• Rules: to make precise and (if desired) temporary modifications to your search results.