BACKGROUND: Triversity offers full infrastructure support for both brick-and-mortar and click-and-mortar retailers in a unified system. Four hundred employees, 350 customers in 32 countries, 25,000 stores, 120,000 terminals and one billion transactions a year are supported by Triversity Inc.
PLATFORMS: K.wiz from thinkAnalytics Corporation is running under Windows NT 4.0 (server and workstations) and attaching to various DBMSs (SQL Server, RedBrick).
PROBLEM SOLVED: Retailers need to understand and manage the effects of customer loyalty programs. As one part of the solutions provided by Triversity, customer loyalty programs can be created, deployed, managed and analyzed. The component provided by K.wiz allows targeted promotions to be created on a one-to-one or individual basis. Advanced searching algorithms allow the individuality of customers represented in their clicks-and-bricks buying behavior and demographic information to be understood and utilized to better create an individual experience for the shopper.
PRODUCT FUNCTIONALITY: The product functions as a back-office number crunching toolkit. Periodic automated analyses of customer transaction data (among other data) builds individualized predictions and summaries of shoppers. This information is presented in the integrated retail reporting environment provided by Triversity and the Allegiance Product Suite. We are proposing to add more automation and extend the individualized analyses of customers' behavior in an online environment by extending the learning and modeling of buyer behavior in the Web-browser arena.
STRENGTHS: The product allows a visual programming representation of complex analyses in a workbench environment, enabling fast development of otherwise difficult analysis. Once created, the visual plan can be managed through a Java API, allowing close control and scripting, such as management of an analysis plan, as it runs. CPU- intensive parts of the analyses (summarizations, modeling, etc.) can be distributed with ease to machines with more resources. The open Java API allows the K.wiz system to be easily tied into our existing and future development infrastructure. The product was evaluated against many others and provided the best combination of analytical components, scalability, open Java API and support. Support allows direct access to developers.
WEAKNESSES: Being a 100 percent Java solution, the GUI is built with Swing technology that, as Sun acknowledges, can be slow in responsiveness. This is a mild irritation and affects only the analyst building the initial plans. Out of the box and with no training, the product has a steeper learning curve than other products. Some of the analytical components are not mature and provide little control; but as the product is evolving rapidly, shortcomings seem to be quickly addressed.
SELECTION CRITERIA: The major reason for selecting K.wiz was the combination of multiple data mining technologies, scalability, a 100 percent Java API and access to quality support. This combination provided a major advantage over all other products investigated.
DELIVERABLES: Due to the very specific requirements and functionality we required, K.wiz produces tables of individualized customer segmentations, recommendations and summarizations that are fed into our retail-specific reporting tools (Allegiance). This separate front end manages all retail- related tasks of loyalty programs, basket analyses, churn analyses, etc.
VENDOR SUPPORT: Quality support was a critical part of our requirements, and thinkAnalytics has not let us down in pre- and post- implementation. The access to knowledgeable developers has been critical in allowing our complex specifications to come to fruition.
DOCUMENTATION: As mentioned, K.wiz has a steeper learning curve than similar products on the market. The documentation is sufficient for most tasks if you are familiar with the rudiments of data mining. Though it lacks a one- page, "what to do straight out of the box" document, a wizard can take you through building an initial plan to build and view a decision tree.
Register or login for access to this item and much more
All Information Management content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access