1.2: Visual Progression of Analytics Types

Figure \(\PageIndex{1}\): A staircase shows four levels of data analysis — from simply describing what happened, to understanding why, predicting what's next, and deciding what to do — each step more complex and valuable than the last. (CC BY 4.0; Felix Amoruwa) Access to a detailed description of Figure \(\PageIndex{1}\)

Descriptive Analytics

Goal: Hindsight - Understanding the past. This type of analytics summarizes what has happened in the past. We can examine historical data to identify patterns, trends, and anomalies, essentially providing a retrospective summary. Descriptive analytics is the most basic form and is common in business reporting. 

E-commerce Example: "Our dashboard shows that sales dropped by 20% last week compared to the previous week. We also see that page views remained steady, but our conversion rate fell from 3.2% to 2.1%." This gives a data-driven picture of past events and is often the first step before deeper analysis.

Diagnostic Analytics

Goal: Insight - Understanding the causes. This type digs into why things happened. Diagnostic analytics tries to explain the causes behind observed outcomes by examining data more deeply and often combining multiple data sources. It involves looking for correlations or patterns that reveal the underlying reasons for a trend. 

E-commerce Example (continued): "We analyzed the timing of the sales drop and discovered it coincided with our recent website update. Further investigation revealed that the new code caused pages to load 40% slower, leading to a significant increase in cart abandonment rates—jumping from 68% to 78%." By identifying root causes, diagnostic analytics helps organizations learn from past successes or failures.

Predictive Analytics

Goal: Foresight - Forecasting the future. This type of analytics uses historical data and statistical or machine learning techniques to forecast what is likely to happen in the future. Predictive models examine historical patterns to predict future outcomes, providing probabilities or trends going forward. 

E-commerce Example (continued): "Based on our analysis of how page speed affects customer behavior, our predictive model forecasts that if we don't fix the slow loading times, we'll likely see a continued 15-20% sales decline over the next month. The model also predicts that customer satisfaction scores will drop by approximately 25%." By anticipating future events, predictive analytics allows businesses to proactively address issues or opportunities.

Prescriptive Analytics

Goal: Action - Recommending optimal decisions. This is the most advanced type, focused on what should be done given predictive insights. Prescriptive analytics suggests optimal actions or decisions by evaluating different scenarios and their likely outcomes. It often incorporates methods from decision science, optimization, and simulation to recommend actionable strategies. 

E-commerce Example (continued): "Our prescriptive analytics system recommends immediately rolling back to the previous website version to stop further sales losses. Additionally, it suggests implementing a new, optimized version that our testing indicates could improve page load times by 25% and potentially increase sales by 5% above the original baseline. The system also recommends sending targeted discount offers to customers who abandoned carts during the slow period to recover approximately 30% of lost revenue." This type of analytics closes the loop by not only anticipating the future but also guiding decision-makers on the best course of action.

Here in Figure 1.2.2 we have a vertical flowchart consisting of four stacked panels connected by downward-pointing arrows, illustrating the four types of analytics applied to a real online store scenario. The first panel, labeled "1. Descriptive — 'What happened?'" with a dark blue header, states that sales dropped 20%, the conversion rate fell from 3.2% to 2.1%, and a dashboard shows a declining revenue trend over the past quarter. The second panel, labeled "2. Diagnostic — 'Why did it happen?'" with a teal header, reveals that page load times were 40% slower after a redesign, cart abandonment rose from 68% to 78%, and the root cause was website performance degradation after a recent update. The third panel, labeled "3. Predictive — 'What will happen?'" also with a teal header, projects a further 15–20% sales decline and a 25% drop in customer   satisfaction if no action is taken, along with continued revenue loss and customer churn. The fourth panel, labeled "4. Prescriptive — 'What should we do?'" with a gold/yellow header, recommends three actions: rollback to the previous website version, deploy an optimized version, and send targeted discounts, with an expected outcome of recovering 85% of lost sales within 30 days. The diagram demonstrates how each analytics type builds progressively on the previous one, moving from identifying a problem to recommending a solution.

Figure \(\PageIndex{2}\): Four panels walk through a real online store problem, showing how each type of analytics builds on the last — from spotting a sales drop to figuring out why, forecasting the damage, and deciding on a fix. (CC BY 4.0; Felix Amoruwa) Access to a detailed description of Figure \(\PageIndex{2}\)

These four types of analytics are often depicted as a progression in value and complexity. Descriptive analytics provides hindsight, diagnostic offers insight into causes, predictive gives foresight, and prescriptive delivers actionable guidance. Together, they enable organizations to move from reactive decision-making to a more proactive and optimized strategy.

The Data-to-Insight Lifecycle

Here in Figure 1.2.3 we have a circular flowchart depicting the seven-step Data-to-Insight Lifecycle as an iterative process. The cycle begins at the top with "1. Define Goals & Requirements," then flows clockwise via directional arrows through "2. Data Collection," "3. Storage & Management," "4. Cleaning & Prep," "5. Analysis & Modeling," "6. Visualization & Communication," and "7. Insight to Action." Each step is displayed as a rounded rectangle with white text on a dark teal background, except for step 7 which uses a lighter teal. From step 7, a dashed orange arrow loops back to step 1, labeled "Iterate & Refine," indicating that the process repeats and improves over time. The center of the diagram is labeled "Iterative Process," reinforcing that the lifecycle is continuous rather than linear. The diagram illustrates the complete end-to-end journey data undergoes — from setting goals and collecting raw data through cleaning, analysis, and visualization, ultimately leading to actionable insights that inform the next cycle. 

Figure \(\PageIndex{3}\): A circular diagram shows the seven steps data goes through — from setting goals to taking action — with an arrow looping back to show that the process repeats and improves over time. (CC BY 4.0; Felix Amoruwa) Access to a detailed description of Figure \(\PageIndex{3}\)

The data-to-insight lifecycle encompasses the end-to-end stages that data undergoes to yield meaningful conclusions.

In practice, the lifecycle spans from the initial collection of raw data all the way to the visual dashboard representation of insights to decision-makers.

Figure \(\PageIndex{4}\): Raw data from many sources flows into a Data Lake for storage, then gets cleaned and organized through an ETL pipeline before landing in a Data Warehouse — where it's finally ready for analysis. (CC BY 4.0; Felix Amoruwa). Access to a detailed description of Figure \(\PageIndex{4}\)

The figure above outlines four core steps in this process: collecting the data, storing the data, cleaning the data, and analyzing the data. In reality, organizations may include additional phases (for example, clearly defining the business problem at the start, or communicating and acting on insights at the end) and often iterate through these steps as new questions and data arise.

The data analytics process follows a structured lifecycle that transforms raw data into actionable insights. While presented as sequential stages for clarity, this process is highly iterative – teams frequently circle back to earlier steps as new information emerges or initial findings prompt deeper questions.