Challenge

We were engaged by a well-known high street fashion retailer to support them in their vision to become a digital leader in an increasingly competitive retail landscape.

The business had selected Endeca, an Oracle-owned software application, to enable search optimisation and online personalisation, to allow their digital channel to respond rapidly to market trends and customer demand.

We were tasked with optimising the performance of the software and offer extensive in-house training to ensure change was sustainable.  We translated business need into actionable deliverables by working with a range of internal and external stakeholders.

In addition to day-to-day optimisation of the software, we were also engaged to plan and implement a specific project to transform the legacy taxonomy infrastructure, using Agile project management.

Approach

Drawing on our solid knowledge of Endeca software, we worked closely with the UX, E-Commerce and Visual Merchandising teams to enable their day-to-day operation and optimisation of the software.

We started by requirement gathering from the key stakeholder teams to understand their knowledge limitations and future aspirations for the system.  We also investigated where the BAU “pain points” were and how these can be systematically addressed.

Once we had drawn up a requirements summary, we established an ongoing, Agile change management process, creating a backlog list and funnelling these work streams into development.  With regular feedback and iterations required, this process worked extremely well as an Agile project.

In addition to the day-to-day software optimisation and internal training activity, we were also co-ordinating the Web Taxonomy Structure project.  Following a full audit of the company’s data model infrastructure, we established that they had a legacy, in-house taxonomy system which used a rigid, category-based hierarchy structure.  In order to maximise the customer search experience we recommended that they move towards a product-based, data enriched model, also known as a so-called “hierarchy-less structure”.

The process started with a Taxonomy “Quick Fix” which used Endeca to differentiate between core and non-core (e.g. promotional) categories that were all held in one place by the legacy system. This allowed a lean and consistent hierarchy structure to be created on site, without the need to overhaul the legacy data.  The second stage was Product Enrichment (tagging) which fundamentally switched the data model from the attributes of a product being based on where it is in the hierarchy, to that information being held against the product record itself. This allowed far more versatility in how the data could be used on-site by the Merchandising teams, whilst leveraging a key strength of the Endeca application: the ability to quickly and dynamically create different navigation states to present to the customer.

There were two additional complexities in this project: the first was the number of internal teams and external agencies who had a stake in the development and optimisation of the website.

The second was that the back-end taxonomy recategorisation was occurring at the same time as in-house front end development, which significantly increased the number of team members working on the website.

Outcome

We successfully trained all team members within the UX, E-Commerce and Visual Merchandising teams to operate the Endeca software on a BAU basis.  In addition we are closely working with key stakeholder teams to identify, define, implement and roll out ongoing improvements to make their website best-in-class.

The Web Taxonomy Transformation Project was successfully implemented to time and budget.  Upon completion the website had a consistent taxonomy structure to present to the customer (and Google!) across the search and browse experience. Back-end business users could now create attributes and add values to products as they are set up.  Data attributed in this way allows users to create collections of products that update dynamically (via Endeca), rather than static categories that need to be updated manually.