标签：Chapter versus Analytics Data customer making data IDA

decision making ≈ data driven decision making

Data-driven decision-making refers to leveraging aggregated and summarized
data to drive critical decisions. The data serves as a compass, allowing you to
refine your “gut feeling” and minimize bias in your decision-making process.
This is in stark contrast to using intuition and observation alone. There are many
reasons for businesses wanting to make data-driven decisions. This can include
optimizing company operations, creating new business opportunities, and
understanding trends in customer behaviors.

Simple versus Intelligent Data Analytics

IDA enhances decision-making by providing deeper insights from complex data sources, beyond what SDA can offer. It requires enriching, standardizing, or transforming data with AI before analysis. Organizations can either build custom AI pipelines, use off-the-shelf tools, or hire data science companies to perform IDA-driven analysis. Ensuring proper data collection and storage is crucial for leveraging IDA effectively.

In the daily operations of a business, various types of data are generated, including:

Sales data
Customer web activity data
Customer emails
Customer call logs
Customer reviews and complaints
Sensor data from manufacturing operations
Quality assurance data
Employee performance reviews

Data Analytics involves summarizing this data with computer assistance to gain insights and track patterns. This can be divided into two categories: Simple Data Analytics (SDA) and Intelligent Data Analytics (IDA).

Simple Data Analytics (SDA)

Simple Data Analytics (SDA) is used to answer straightforward questions using a subset of data. For example, calculating Monthly Recurring Revenue (MRR) involves summing the monthly fees paid by every customer without manipulating the data. SDA is effective for basic questions related to sales and profit. It is a method that most businesses use today to perform basic calculations and derive insights from structured data.

For instance, to calculate MRR, you access the customer payment database and sum the monthly payments for all customers. This straightforward approach does not require data manipulation or transformation and is typical of SDA.

Intelligent Data Analytics (IDA)

Intelligent Data Analytics (IDA), on the other hand, adds a layer of intelligence to the data, often using Machine Learning (ML) and Natural Language Processing (NLP) to gain deeper insights. IDA is particularly useful for messy, semi-structured, or unstructured data. It can:

Add information not present in raw data
Summarize large amounts of unstructured data
Reduce manual and time-consuming work

For example, to understand customer demographics, IDA can predict customer attributes using ML models, providing a comprehensive overview of customer profiles. This approach allows businesses to answer more complex and non-obvious questions, such as identifying customer segments or understanding customer behavior patterns.
Example of answering “Who are our customers?” with IDA.

Benefits and Applications of IDA

New Product Innovation

By monitoring customer pain points online, companies can discover hidden product opportunities. For instance, Ocean Spray used NLP to analyze thousands of online conversations about cranberry juice. This analysis revealed unexpected customer behaviors and preferences, leading to the launch of new beverage lines like Ocean Spray Mocktails, which catered to customers using cranberry juice as a non-alcoholic drink substitute.

Improving Customer Experience

Leveraging disparate data sources like social media, support emails, and customer reviews, companies can centralize and analyze feedback to enhance products and services. For example, a hospital used IDA to analyze web comments about their services.** By performing sentiment analysis and extracting discussion themes, they identified key areas for improvement**, such as staff behavior and wait times, leading to actionable insights for enhancing patient experience.

Understanding Employees

Open-ended feedback from employee surveys can be analyzed using NLP to standardize responses and identify key issues. This helps in understanding employee concerns and improving workplace conditions. For instance, analyzing responses to questions like "How can we make this a better place to work?" can reveal common themes such as the need for better pay or a safer work environment, which can then be addressed by management.
To help with this analysis, we used NLP to standardize and simplify the responses. This made it easier to make sense of what employees were saying and understand which issues needed more attention than others.

Visualization of free-form responses to the question “How can we make this a better place to work?” standardized using NLP. The data shows the number of respondents wanting “more pay” segmented by location.

Enhancing Search and Marketing

Analyzing search logs from site search engines can provide insights into user behavior and improve search functionality, content strategy, and SEO. For example, search logs can reveal the most common search queries, the effectiveness of search results, and user engagement with the search functionality. This data can be used to refine the search engine, improve content discoverability, and enhance the overall user experience on the website.

Automating Root Cause Analysis

In manufacturing and healthcare, IDA can cluster incident reports to identify prevalent issues and their root causes, facilitating corrective actions. For instance, in cancer radiation therapy, incident learning systems track errors during patient treatment. By clustering these incidents using ML, stakeholders can identify common issues such as "wrong patient positioning" or "human error" and take corrective measures to prevent recurrence.
An example of problem clusters in cancer radiation therapy. A drill-down of each cluster can reveal the root causes.

Integrating with Business Intelligence Tools

IDA can augment BI tools to answer complex questions from unstructured data sources, (you need to augment the tools with distinct NLP and ML pipelines before analysis), providing deeper insights for decision-making. For example, integrating ML and NLP pipelines with BI tools can help surface top customer complaints from social media or group customers based on their behavior patterns, enabling more informed business decisions.
Example of how IDA fits into existing BI tools. This example uses a sentiment analysis pipeline. In practice, there can be multiple AI pipelines to produce different types of AI-driven enrichment, standardization, and summarization.

Optimizing Decision-Making with IDA

To leverage IDA effectively:

Identify Data Sources: Understand the complex data sources available in your organization, such as customer interactions, operational data, and external data sources.
Define Questions: Determine the types of questions you want to answer using these data sources. For example, you may want to understand customer preferences, identify operational inefficiencies, or predict market trends.
Build AI Pipelines: Develop custom AI pipelines or use off-the-shelf solutions to enrich and analyze data. These pipelines may involve data preprocessing, feature extraction, model training, and result interpretation.
Ensure Data Readiness: Streamline data collection, storage, and access to facilitate IDA. This involves ensuring data quality, integrating disparate data sources, and maintaining data privacy and security.
Conclusion