Applying Aster’s Capabilities > Search Result Optimization > Behavioral and Predictive Analytics

Teradata Guided Analytics
Teradata Employee

Aster as an analytic platform with multiple capabilities is a powerful business tool if you harness and apply it to high impact business problems. We have multiple examples that could be used to demonstrate this, but in this post I will focus on just one of the Solutions we have built in the particular arena of online retail.


One of the greatest advantages of an online presence is the ability to offer an almost unlimited number of choices to your visitors. But this reality also comes with its own set of challenges: how do you feature and present the right products to your prospects? There are certainly clues provided during the visit, the strongest of which is a site search for the product or service of interest. The user types in a search term and you have one shot to discern what exactly s/he is looking for from your vast inventory of potential matches.


One approach is to simply return an infinite scroll of matches. This might be acceptable for some desktop searches, but consider that the majority of ecommerce revenue has shifted to mobile platforms with limited screen real estate and fewer navigation aids (such as search refining and filtering).


Our approach to onsite search optimization leverages multiple Aster capabilities to maximize the relevance of search results your site returns in response to a search term.


One component of our approach can be classified as text analytics in which the natural language product descriptions for the entire inventory are ingested by our machine learning algorithms to identify the most relevant keywords for each and every product. This process automates the generation of keywords that your indexing and presentation system uses to prioritize the most relevant results above others. The result is a predictive model that maximizes the relevancy of any search action performed on your ecommerce site.


Simultaneously, we employ Aster’s capability to perform behavioral analytics to ingest and analyze event logs of recorded visitor actions. We could just look at the pathways that end in conversion, but while this may be the ultimate goal, events other than conversion such as a:


  • site exit
  • drill down to more detail
  • refinement or filter of results
  • follow-on search(es)


all contain clues about what your customers are/are not looking for and should be mined to fine tune future search results.


This approach to onsite search optimization has yielded enviable results at a national online retailer. In direct A/B testing these incremental optimization efforts yielded a 2% lift in conversion rates and a 3.5% increase in revenue per visit with a corresponding increase in average order volume.


If you want to discuss this instance of multi genre analytics and the significant business impact it can deliver please get in touch.