Elevate Your Search Experience by integrating Coveo into AEM Cloud implementation - Part 2
Part 2: Practical Example of 5-Steps Coveo implementation
This is my second blog post in this series. This is continuation of my first blog in this series: Part 1: Launch Your Coveo Search Exploration
In this blog, I will take a closer look at the Coveo-AEM implementation in which I was involved. I will provide a thorough examination of the five-step Coveo implementation process, offering insights into the outcomes at each stage.
In my recent involvement in the replatforming journey of a significant client transitioning to AEM Cloud(Adobe Experience Manager Cloud platform), a pivotal aspect of the project revolved around enhancing findability within their existing application.
For this particular implementation, the search system was expected to meet the following key requirements:
Personalized Search Experience: Tailoring the search functionality to provide a personalized experience for logged-in users.
Secure Search Experience: Ensuring the security of the search process by delivering results based on user roles, thus restricting access appropriately.
After an extensive initial discovery phase, the decision was made to implement Coveo Search as a composable solution seamlessly integrated with AEM. This strategic choice was driven by the platform's ability to fulfill the specific requirements for personalized and secure search experiences.
The subsequent steps involved conducting in-depth workshops with key stakeholders to outline the details of the Coveo implementation. Following the outlined 5-step process, detailed in part 1 of this blog series, I thoroughly explored each aspect to ensure a robust and tailored search solution.
Outcomes achieved:
Step1 : Understanding End Users and Expectations:
The outcome of this step resulted from workshops conducted with business users, delving into the details about the end users and what a delightful search experience means to them.
Identified diverse end-user categories, including anonymous and logged-in users with various roles.
Role-based access restrictions for logged-in users, ensuring they could only search for items associated with their designated roles.
Result refinement based on identified tags/metadata.
Defined relevance rules to provide a personalized search experience.
Step2 : Data Repository and Coveo Connector Identification:
After understanding the end users and their key expectations, the next step involved examining the content tree structure within AEM, including content pages and assets. We also conducted a joint workshop with the technical and business teams to define the content tagging structure to be used by content authors.
Established that all necessary content and assets for search would reside in AEM.
Defined the content tagging structure for content and assets sitting in CMS.(AEM Cloud)
Selected the Coveo generic Sitemap connector, aligning it with our requirements and AEM's proposed content structure.
The sitemap format was determined based on the proposed content structure in AEM, which could be integrated with the Coveo connector. I'll share a snippet of it in a future blog post in this series, where I'll discuss the details related to the Sitemap connector.
Step 3 : Search Experience Design
This step involved creating prototypes for the user search journey and identifying the Coveo UI framework to be used for development.
I crafted prototypes utilizing Coveo's search interface builder to promptly showcase the array of available Coveo search components. Subsequently, these prototypes were shared with the UX team to inform the proposed designs.
Upon approval of the design, the decision was made to leverage the Coveo Atomic UI framework. This choice was driven by considerations such as performance requirements, project timelines, and the desired level of customization.
Figured out how to blend Coveo Atomic components seamlessly into the AEM search page. I will be covering details of the component integration in a future blog post in this series.
Step 4 : Search Relevance
This was achieved through workshops with business stakeholders to determine the Coveo query pipeline rules and AI models that could be used to provide the desired search experience for end users.
Leveraged existing search analytics data to identify an initial list of synonyms/thesaurus rules.
Proposed a phased rollout of Coveo Machine learming models, including Automatic Relevance Tuning (ART) and Query Suggestion (QS) models in the first phase.
Automatic Relevance Tuning (ART) model. - An Automatic Relevance Tuning (ART) model enhances the ranking scores of items often opened by users with similar searches and roles. This adjustment is based on the current search query, ensuring more relevant results for users in similar contexts.
Query Suggestion (QS) models - this models can provide search term (query) completion suggestions that have showed relevant results in the past based on the user's role.
Step 5 : Continuous improvement
This step was consistently part of our workshop discussions, focusing on ensuring we have the appropriate resources and skillsets identified from the business side to support this step after going live.
Identified resources to collaborate with Coveo's customer success manager, utilizing Coveo usage analytics dashboards for continuous review and relevance tuning.
Proposals for having customized Coveo analytics dahboard for the customer.
Reveiwing options to export Coveo usgage analytics data to external systems using Coveo analytics APIs.
In conclusion, this blog has covered the detailed steps I took to integrate Coveo with AEM Cloud. It's not just about addressing current needs; it's about establishing a foundation for continuous enhancements.
Stay tuned for the next part of this series, where I will cover details on how we used the Coveo Sitemap connector for this implementation.