SELECT
s.CompanyName AS Supplier,
s.Country,
COUNT(p.ProductID) AS ProductCount
FROM Suppliers AS s
JOIN Products AS p
ON s.SupplierID = p.SupplierID
WHERE p.Discontinued = 0
GROUP BY
s.CompanyName,
s.Country
ORDER BY ProductCount DESC
LIMIT 10;Analysis Report Two - Systems Integration and Decision Support
Executive Summary
Healthcare organizations depend on connected information systems to manage daily work and make larger strategic decisions. Transaction Processing Systems collect routine records such as patient registrations, payments, supply movements, and medication transactions. Management Information Systems turn those records into useful reports. Enterprise Resource Planning systems connect information across areas such as finance, human resources, purchasing, inventory, and operations. When these systems work together, employees can receive faster and more complete information. However, more information does not always lead to better decisions.
The assigned readings show that successful integration depends on data quality, employee training, clear goals, and strong governance. Harvard Business Review Analytic Services found that 87 percent of surveyed executives believed their organizations would be more successful when frontline workers could make important decisions in the moment. At the same time, 86 percent said those workers needed better technology-enabled insight (Harvard Business Review Analytic Services 2020). This means that access to information and the authority to act must develop together.
The Northwind analysis in this report demonstrates how SQL joins can turn separate transaction tables into focused management information. The first visualization connects suppliers and products to show supplier concentration. The second connects employees, orders, and order details to compare the revenue associated with orders handled by each employee. Although Northwind is a sales database, the same method could be used in healthcare to connect suppliers with medical products or connect staff, patient encounters, and charges.
The main recommendation is that organizations should begin with a specific business question instead of beginning with a technology. They should improve data quality, limit reports to useful measures, train employees to interpret results, and establish security and decision rights. These steps allow systems integration to support strategy without creating information overload.
Introduction
Information systems form a hierarchy that supports different levels of work and decision-making. A Transaction Processing System, or TPS, operates at the daily level. It records structured and repetitive events such as purchases, payments, inventory updates, patient registrations, and pharmacy transactions. These systems must be fast, accurate, and reliable because a failure can interrupt normal operations. A point-of-sale register is a common example, while a hospital registration or medication administration system provides a healthcare example (Church 2026).
A Management Information System, or MIS, takes the detailed records created by a TPS and summarizes them into information that managers can understand. Instead of reading thousands of individual transactions, a manager might review a report showing revenue, inventory levels, patient volume, staffing needs, or changes over time. These reports support structured and semi-structured decisions by making patterns easier to see.
Enterprise Resource Planning, or ERP, expands integration across the organization. An ERP system can connect finance, purchasing, inventory, human resources, payroll, scheduling, and other functions through a shared database. When one department updates a record, the information becomes available to other authorized users. In healthcare, an integrated system can help leaders connect staffing costs, supply use, patient volume, and financial performance. It can also reduce duplicate data entry and delays between departments.
The New Decision Makers explains why useful information should not remain only with executives. Frontline employees are often closest to customers, patients, equipment, and daily operations. The report found that only about one-fifth of surveyed organizations had frontline workers who were both empowered and digitally equipped to make decisions effectively (Harvard Business Review Analytic Services 2020). Better access to information can help employees solve problems faster, but the information must be understandable and connected to their responsibilities.
The assigned material about muddled predictions adds an important warning. More data can make a decision worse when additional variables are weak, biased, irrelevant, or difficult to interpret. Research by Csaszar, Jue-Rajasingh, and Jensen found that using all available information does not always improve prediction accuracy. In some situations, fewer and more relevant factors produce a better and fairer result (Csaszar, Jue-Rajasingh, and Jensen 2022).
The main argument of this report is that TPS, MIS, and ERP systems improve organizational decision-making when they connect reliable operational data to a clear business need. Integration can hurt the organization when poor data spreads across departments, employees receive too many measures, or leaders adopt technology without preparing the workforce to use it.
Outside Research
Joe Peppard argues that most organizations should not create a separate artificial intelligence strategy before they complete the basic work needed to use AI successfully (Peppard 2025). His argument connects directly to systems integration because AI depends on existing data, infrastructure, workflows, and employee knowledge. A company cannot create a dependable AI system from incomplete, biased, inconsistent, or unstructured data.
Peppard uses predictive maintenance as an example. A manufacturer may want AI to predict when a machine will fail, but the organization first needs historical records showing previous failures and their warning signs. It also needs sensors to collect current information, connectivity to transfer the data, cloud infrastructure to store it, and a workflow that sends useful alerts to employees. AI is only one part of the complete system.
A similar situation can occur in healthcare. A hospital may want to predict when imaging equipment will need maintenance. The hospital would need accurate equipment histories, repair records, sensor data, scheduling information, and a process for assigning the repair. If those systems are not integrated, the prediction may arrive too late or reach the wrong employee.
Peppard’s argument also shows why organizations should begin with a business problem instead of a popular technology. Asking how a hospital can use AI may cause leaders to treat AI as the solution to every issue. A better process is to identify the problem first, such as equipment downtime, supply shortages, scheduling delays, or claim denials. Leaders can then decide whether the best solution is AI, a basic MIS report, a database improvement, an automated workflow, or another technology.
The research about using less data supports this approach. Csaszar and his coauthors found that adding information can reduce predictive accuracy when the information is applied inconsistently or has a weak connection to the outcome (Csaszar, Jue-Rajasingh, and Jensen 2022). This does not mean organizations should ignore data. It means they should choose information based on the decision rather than assuming that every available variable improves the result.
Employee preparation is another major issue. The New Decision Makers found that 68 percent of surveyed organizations trained frontline employees to use technology tools, but only 46 percent trained them to apply the insights produced by those tools (Harvard Business Review Analytic Services 2020). An employee may know how to open a dashboard but still not understand which measure matters, what the graph leaves out, or when the result should be questioned.
Technology integration can therefore help or hinder strategic goals. It helps when it reduces delays, connects departments, improves visibility, and gives employees focused information. It hurts strategy when poor data spreads through connected systems, dashboards create overload, or leaders choose a technology before defining the problem.
Data Visualizations
The Northwind database represents operational data that could be collected by a TPS. Its tables store separate records about suppliers, products, employees, orders, customers, and individual items within orders. SQL joins connect those records so they can be summarized for management use.
Visualization One - Two Table Join
The first query joins the Suppliers and Products tables through SupplierID. It counts the number of active products provided by each supplier and displays the ten suppliers connected to the largest product assortments. This can help managers identify whether the organization depends heavily on a small number of suppliers.
ggplot(
data = myquery1,
aes(
x = reorder(Supplier, ProductCount),
y = ProductCount
)
) +
geom_col(
fill = "#4F7C82"
) +
geom_text(
aes(label = ProductCount),
hjust = 1.2,
color = "white",
size = 3.5
) +
coord_flip() +
theme_minimal() +
labs(
title = "Suppliers Supporting the Largest Product Assortments",
subtitle = "Two-table join between Suppliers and Products",
x = "Supplier",
y = "Number of Active Products",
caption = "Source: Northwind SQLite database"
)
This graph transforms separate supplier and product records into a focused supply-chain measure. Managers can quickly see which suppliers are connected to the largest numbers of active products. However, product count alone does not measure the full importance or risk of a supplier. A supplier may provide fewer products that generate more revenue or support an important product category.
Important
For the required live edit, the Categories table will be added to this query. This will change the two-table join into a three-table join and show what types of products each supplier provides.
Visualization Two - Three Table Join
The second query joins the Employees, Orders, and Order Details tables. It calculates the sales revenue connected to orders handled by each employee. Revenue is calculated using the unit price, quantity, and discount for each line item.
SELECT
e.FirstName || ' ' || e.LastName AS Employee,
COUNT(DISTINCT o.OrderID) AS OrdersHandled,
ROUND(
SUM(
od.UnitPrice *
od.Quantity *
(1 - od.Discount)
),
2
) AS Revenue
FROM Employees AS e
JOIN Orders AS o
ON e.EmployeeID = o.EmployeeID
JOIN "Order Details" AS od
ON o.OrderID = od.OrderID
GROUP BY
e.FirstName,
e.LastName
ORDER BY Revenue DESC;ggplot(
data = myquery2,
aes(
x = reorder(Employee, Revenue),
y = Revenue
)
) +
geom_col(
fill = "#D6A84B"
) +
coord_flip() +
scale_y_continuous(
labels = scales::dollar_format()
) +
theme_minimal() +
labs(
title = "Revenue Connected to Orders Handled by Employee",
subtitle = "Three-table join using Employees, Orders, and Order Details",
x = "Employee",
y = "Sales Revenue",
caption = "Source: Northwind SQLite database"
)
This visualization shows how systems integration can support employee and sales analysis. A manager can quickly see whether revenue is distributed evenly or concentrated among a few employees.
However, revenue alone does not fully explain employee performance. The results do not account for territory size, customer assignments, product mix, employee workload, returns, or customer satisfaction. Managers should use the graph to identify patterns and ask additional questions rather than treating it as a final employee ranking.
Recommendations for Industry
Start with a clear business question
Organizations should define the problem they are trying to solve before selecting a technology. Leaders should identify who will use the information, what decision must be made, and what result would demonstrate improvement.
Improve the data foundation
Integrated systems need consistent definitions, accurate records, ownership standards, and correction procedures. A shared database can improve coordination, but it can also spread inaccurate information quickly.
Limit information to what is useful
Employees should not receive every available measure. Reports and dashboards should be designed around the employee’s job and decision-making responsibilities. Focused information reduces overload and makes the system easier to understand.
Train employees to apply insights
Training should include more than instructions about how to operate the technology. Employees also need to understand what the information means, what it leaves out, and when human judgment is necessary.
Balance empowerment with governance
Frontline employees need enough information and authority to respond quickly. However, organizations must also establish access controls, approval limits, privacy protections, and accountability.
Treat AI as part of a larger system
AI should be treated as one possible part of a broader digital strategy. Before investing in AI, leaders should evaluate data quality, system compatibility, employee readiness, security, and expected business value.
References
Church, Mitchell. 2026. “Topic 2 Overview: Systems Integration and Decision Support.” Course video for CBAD 393 Management Information Systems, Coastal Carolina University.
Csaszar, Felipe A., Diana Jue-Rajasingh, and Michael Jensen. 2022. “When Less Is More: How Statistical Discrimination Can Decrease Predictive Accuracy.” Organization Science 34 (4): 1383–99. https://doi.org/10.1287/orsc.2022.1626.
Harvard Business Review Analytic Services. 2020. “The New Decision Makers: Equipping Frontline Workers for Success.” Research Report. Harvard Business Review Analytic Services.
Peppard, Joe. 2025. “Why Most Companies Shouldn’t Have an AI Strategy.” The Wall Street Journal.