Big data has revolutionized business, allowing organizations to turn troves of information into meaningful insights. Yet with so many different data sources, information often becomes siloed and business leaders lack the visibility needed to make data-driven decisions at the enterprise level. Most of these stores of information get lost in different tools and databases, and switching between these different programs to gather the right data becomes too time-consuming. To achieve repeatable processes and gain valuable insights across business processes, today’s enterprises need a highly-scalable System of Insight.
A System of Insight generally refers to technology that includes three fundamental components:
- A system of record, which gathers, connects, and stores structured enterprise data
- A system of engagement, which connects people, things, social elements, and unstructured or textual data
- A system of automation, which allows a cycle of improvement.
A System of Insight can improve decision-making by allowing users to uncover patterns and anomalies in their enterprise data. This powerful framework doesn’t just unlock the findings that matter most to a business – it also ensures the entire enterprise can access them, enabling results to be continuously measured and improved.
OpenText™ Magellan™ operates as a System of Insight in a single platform with four steps: signal, action, outcome, and improvement. Each step builds upon the others, which allows for incremental improvements that are then used to fine-tune the signal and create a continuous improvement cycle. Following the four-step model, here’s how you can use Magellan as a System of Insight.
1. Signal: Use interactive dashboards to explore data and discover useful insights.
OpenText Magellan allows users to uncover trends and useful insights through interactive visualizations. One example of a user who might use this module regularly is a marketing analyst in a wealth management firm. To identify which product categories are performing well versus which may need more attention, the analyst can look at both an overview of performance in a single line chart, or drill down to further explore a key area, such as investments.
For example, upon taking a closer look at investments, the analyst might discover that while most products are climbing, investments in individual savings accounts (ISA) are declining. Because Magellan compiles data from a broad range of sources, the analyst can compare the relative performance of investments in ISA to the performance of competitors’, and discover that other firms are actually excelling in this area.
2. Action: Initiate action based on an analysis of data.
Based on these findings, the marketing analyst wants to run a campaign to recommend investment ISA products – but not to all customers. The firm’s data scientist can easily build a hyper-optimized recommendation model using Magellan’s data science notebook, a Jupyter-based environment that allows data scientists to pick up any data they like and build a collaborative filtering model. They can train, test, and finalize the model by creating an algorithm and putting it into a pipeline.
Because components of Magellan are tightly integrated, data scientists can create a machine learning model that can be published and made available to the rest of the Magellan platform. Now, using Magellan Data Discovery, other analysts can access the model in the machine learning library and drag a customer list into the recommendation model to see the likelihood of the customer opening an investment ISA. With this information, they can initiate a marketing campaign directly from the Magellan dashboard and check the outcome of the campaign once complete.
3. Outcome: Track and analyze the performance of our actions to produce a tangible outcome.
If the marketing analyst finds the marketing campaign didn’t perform as well as hoped, they can go back and drill down for further investigation using Data Discovery to explore key analytics for existing customers. The analyst can create a profile of customers who have investment ISA products by simply dragging in data points like age, income level, homeownership, and other demographic information.
A Z-score for each attribute is automatically created and sorted to visually indicate the attributes most statistically likely for a customer. Analysts can then compare the marketing spend against the actual customer demographic. They might find that although most customers who have investment ISA products are over the age of 75, the company has recently decreased their spend for this segment. With these new findings, the data scientist can create a new machine learning model to improve segmentation based on these demographics.
4. Improvement: Use the results of the analysis to improve the signal or process, thereby completing a full System of Insight cycle.
Clearly, the analyst saw that the outcome of the first marketing campaign didn’t meet expectations, which prompted revisiting the data. With a new machine learning model based on more granular customer insights, an analyst can now create a more targeted campaign, and can even schedule a workflow that triggers the campaign to run on a set schedule.
Once again, the analyst will track the outcome and use those findings to further refine the process where needed or tackle another product category, thereby creating a full System of Insight cycle.
While the System of Insight cycle is an iterative process that requires some fine-tuning, each step allows different users to uncover new insights about the organization. Findings only improve over time, making predictions more accurate to help inform data-driven decisions and accelerate business.