Digital Developer Conference: Hybrid Cloud. On Sep 22 & 24, start your journey to OpenShift certification. Free registration

Analyze unstructured data with AI to gain product performance analysis

Summary

To leverage the voice of the customer to drive business results, companies need AI to gain actionable insights from sentiment, emotion, concepts, and keywords mentioned in customer feedback. Much of the important data a customer wants to share can already be found. It exists in public forums, blogs, social media posts, and chat logs with customer representatives. The challenges to leveraging this information have historically been the unstructured nature of it. To address this, IBM has AI solutions such as services like IBM Watson® Discovery that can be trained to aggregate, enrich, and help surface key customer insights.

In code patterns for this solution, unstructured data consisting of product reviews and customer surveys are imported into Cognos® Analytics from IBM Watson® Discovery.

Description

Cognos Analytics is a business intelligence solution that empowers users with AI-infused self-service capabilities that accelerate data preparation, analysis, and report creation. IBM Watson® Discovery is an insight engine and AI search technology that breaks open data silos and retrieves specific answers to your questions while analyzing trends and relationships buried in your customer care data. Watson Discovery applies the latest breakthroughs in machine learning, including natural language processing capabilities, and is easily trained on the language of your domain. In this solution, unstructured data consisting of product reviews and customer surveys is imported into Cognos Analytics from Watson Discovery, and Watson Discovery identifies the sentiment toward the products surveyed. This data can then be displayed on a Cognos Analytics dashboard along with corresponding sales revenue and product inventory data. By combining Cognos Analytics with Watson Discovery, you can:

  • Get early warning of trends based on customer feedback
  • Identify and address escalating customer interactions with early intervention
  • Understand customer preferences to target the right customers with the right products and the right marketing
  • Know what is driving growth and where to invest in your business to drive revenue and customer adoption
  • Catch product inventory issues early for trending products before they take off
  • Get to the root cause of churning customers
  • Identify actions to increase your net promoter score or other customer satisfaction metric

The following image gives an example of the types of visualizations that are created in these code patterns.

Dashboards

Architecture flow

Architecture

  1. Product review data is loaded into Watson Discovery for enrichment. Results include sentiment analysis and keyword discovery.
  2. Optionally: Product and business data is loaded into many different databases, such as IBM Db2®, Netezza® Performance Server, Mongo Db, MySQL, and so on.
  3. User runs Cognos Analytics.
  4. Cognos Analytics can be linked to databases like IBM Db2, Netezza Performance Server, Mongo Db, MySQL, or data files load directly into Cognos Analytics.

The theme of these code patterns is built around data for a small coffee manufacturer that sells products in local markets. The data consists of reviews and ratings for the different coffee flavors, as well as associated sales and inventory data. We’ll cover how to incorporate data from multiple sources and how to create the visualizations in Cognos Analytics to best represent the data.

The code patterns in this solution explain how to combine Cognos Analytics with Watson Discovery to get early warning of customer feedback trends and how to visualize business data like store information, sales data, and inventory levels in Cognos Analytics for product performance analysis.

Visualize unstructured data from Watson Discovery in the Cognos Analytics Dashboard

db-1-final-dashboard

In the Visualize unstructured data from Watson Discovery in the Cognos Analytics Dashboard code pattern, you learn the steps to:

  • Gather product review data.
  • Upload data into Watson Discovery for enrichment and analysis.
  • Query Watson Discovery to capture keywords and sentiment.
  • Upload data into Cognos Analytics to create data modules and dashboard visualizations.

Visualize customer insights with business data for product performance analysis

db-2-final-dashboard

The Visualize customer insights with business data for product performance analysis code pattern builds on the first code pattern by:

  • Adding sales, store, and inventory data to support our coffee company product data
  • Uploading all product and business data into databases like IBM Db2, Netezza Performance Server, Mongo Db, MySQL
  • Connecting Db2 Warehouse instance, Netezza Performance Server, or Mongo Db or any other that Cognos Analytics supports, to Cognos Analytics
  • Creating additional Cognos Analytics dashboards to visualize business data