The Typical Ipiphany Insight Discovery Process

1. Define your objectives

Like any analysis, cognitive analytics should be performed with clear objectives in mind. Before you begin any analysis, you should reflect on business objectives and requirements to determine what it is that you are trying to learn, the decisions you needing to make or the actions you are wanting to take.

2. Source data

Determine what data you need or have that may contain the information required. Depending on your business objectives and resources available this will likely involve

Structured data – all the same data you would use in traditional analysis projects can be used. Unlike conventional regression analysis Ipiphany supports both multivalued data fields and hundreds of variables  - so there is no need to discard data at this stage that may potentially be combined with unstructured data to help discover key insights

Unstructured data – a key strength of Ipiphany Cognitive Analytics is that it can extract information and meaning from free text verbatim data that would normally be ignored in conventional analysis projects. So you should look hard for potential sources of this data that would normally be overlooked. Examples include Survey Comments, CRM Notes, Social Media Posts and Comments, Customer Complaints Notes and Voice and Chat Transcriptions.

Combining data from multiple sources – the search for new unstructured data sources will likely result in the need to combine data from different sources. Although Ipiphany will not automatically do this for you, this can be achieved in the data preparation stage – so don’t stop at just one data source.

3. Data understanding and preparation

• Familiarise yourself with your data – identify any data quality issues and what potential information is contained within the fields
• Prepare your data – including merging data from different data sources and generating summary and derivative fields to support the discovery process. Ipiphany Consultants are available to support and advise you as required.
• Import a dataset including: Structured fields, Key metric fields (relating to your the objectives), One or more unstructured text verbatim fields

4. Discovery

 In this phase, you will be driving an iterative process which consists of discovering and validating insights by:

• Defining analysis queries that instruct Ipiphany to identify the most important patterns that impact your target business metrics 
• Review and explore the resulting patterns using the linked verbatim aspects and examples 
• Use your knowledge of the business and customerto identify, understand and validate the root causes and relationships between them 
• Develop and refine your Insight Hypotheses - then quantify and validate them by running refined analysis queries to measure the impact on your business metrics and extract examples of verbatim that best illustrate or explain the root causes

5. Evaluation

At this stage in the project you have developed one or more insights that appear to have both high quality, from a data analysis perspective, and are easily understood and validated from a business perspective. Before proceeding to final deployment of the insights, it is important to more thoroughly evaluate and validate the insights against business and domain knowledge, and to be certain they properly achieve the business objectives. A key objective is to determine if there is some important business issue that has not been sufficiently considered. At the end of this phase, a decision on the use of the insights should be reached.

6. Deployment and Action

Creation of the Insight hypotheses is generally not the end of the project. Even if the purpose of the project is to increase knowledge of the data, the knowledge gained will need to be organised and presented in a way that is useful to the business. Depending on the requirements, the deployment phase can be as simple as generating a report or as complex as implementing a repeatable data scoring (e.g. segment allocation) or data mining process. In many cases it will be the business, not the analyst, who will carry out the deployment steps – so presentation of results in a way that includes both statistical measures and relevant verbatim examples is key to facilitating the transfer of insight.