A Comprehensive Guide to Operational Analytics

8 Minute Read

operational analytics

According to a Splunk study, 81% of business and IT decision-makers believe data is “extremely” or “very” crucial to their company’s success. Despite this, they fail to make the most of existing data. The same study disclosed that 55% of their data is untapped.

Untapped data is captured daily as an output of users’ daily actions with various systems and devices. Examples include detailed notes of customers’ sentiments and thoughts. This type of data is collected and stored, but it’s often not used for analytics. Despite being overlooked, this data can offer a tremendous amount of value.

With operational analytics, you can tap into this data to extract valuable insights. It can build on your existing analytics by embedding analytics into your core business functions. It uses automation to make your processes smarter and reduce dependence on data analytics specialists.

What Is Operational Analytics?

Operational analytics is a subset of business analytics that focuses on improving daily operations within an organization.

Consider a hospital administration that wants to improve their daily operations by tackling the following questions:

  • How many nurses are needed for all shifts?
  • How much time should it take to transfer a patient to the ICU?
  • How can we reduce the patient waiting time?

Operational analytics can be utilized to answer these questions and generate insights that improve efficiency and throughput.

Operational analytics doesn’t focus on supporting big or strategic decisions that align more with business intelligence (BI). Instead, it influences the small and tactical decisions that are made daily. It focuses on sending data to the tools that business users utilize regularly. A marketer doesn’t have to open a BI tool to find out about their target audience’s activity on social media. Instead, they can find the information right in their primary tools (e.g., HubSpot).

Operational analytics also ensures real-time data delivery. For instance, an e-commerce company can use operational analytics to collect real-time data about their customers. If users suddenly start displaying unusual behavior, such as abandoning their cart at the last minute after adding multiple items, then the company can review the website’s client-side to see if there’s a buggy website component (e.g., a faulty dropdown menu) that’s preventing users from buying.

How Are Models Developed in Operational Analytics?

Analytic models help you understand data, generate predictions, and make business decisions. There are three primary ways to build models in operational analytics.

1. Model development by analytic professionals

One of the common ways to develop models is to let analytic specialists — usually statisticians or data scientists — build them. They perform quantitative or data analysis through a number of techniques, such as cluster analysis, cohort analysis, regression analysis, etc.

These models can be created in two ways: One, a specialized modeling tool is used. This tool enables the user to access, clean, aggregate, and analyze data. Another way to create a model is to use a scripting language that has substantial support for statistical and quantitative operators. Two of the languages that fit the bill are Python and R.

2. Model development by business analysts

Some organizations only need to make a few analytic models and can’t justify the investment to hire an analytic specialist. They find it more feasible and practical to hire a business analyst. This is usually someone who knows the following:

  • How the business works
  • What types of data are collected
  • How to use business intelligence (BI) tools for reporting

Business analysts generally lack the statistical and advanced programming knowledge of analytic specialists. They rely more on automation to build models by using tools that offer built-in functionalities. Some of the popular tools used by business analysts for modeling include Power BI and Tableau. When needed, they can move data from a database or a data management environment to their tools where they can analyze it.

3. Automated development

The third approach is to build the models automatically using software. You have to define constraints for your decision, figure out how the software will make a good decision, and use the software to develop a model via experimentation. Experimentation here means having the model use multiple approaches for different customers and finding the approach that performs the best. With time, the software gains an understanding of what the customer prefers and creates a model that follows it.

How Can You Implement Operational Analytics?

Certain technologies or processes can help you improve your implementation of operational analytics.

Get the required tools

To ensure you can implement operational analytics, you need to have the right set of tools. You will need a powerful ETL tool to integrate data from a wide range of sources. These sources include operational systems like enterprise resource planning (ERP) software, customer relationship management (CRM) systems, and other systems that help you collect and present data on your organization’s interactions.

You also need BI platforms, specialized tools for data modeling, and a repository like a data lake or data warehouse.

Opt for in-memory technologies

Traditional business intelligence (BI) tasks depend on the tables of a database stored on a disk. With a reduction in memory costs, it’s more practical and faster to access RAM. You can use in-memory technologies for operational analytics to get data access that can improve application performance by minimizing latency bottlenecks.

You can load data into a memory pool of considerable size and run whole algorithms within the memory. This approach is useful when you need to develop and redevelop several complex models on a regular basis. In-memory technologies can be especially useful for building risk models in large-scale financial firms and institutions. A financial firm updates risk models daily for hundreds of securities and scenarios. That’s how they determine how to invest or hedge.

Implement decision services

A viable approach to embed operational analytics is to build decision services. A decision service is a callable service that is used to answer questions or make decisions. A decision service isolates the business logic behind your decisions, separates it from the processes, and reuses it across several applications in various operational environments. This way, it can automate your decision-making and save resources like time and money.

For example, you can set the rules of an insurance claim in a decision service. It can help customers find out if they have a valid claim before they file an auto insurance claim. This service can also be used for other types of insurance.

To come to the right decision, a decision service can use predictive analytics, optimization technologies, and business rules. However, it’s important to ensure that the service can access all existing components of the data infrastructure, such as a data warehouse, database, or BI visualization.

Share a data definition among your teams

You have to make sure that your three core groups — analytics, IT, and business — share the same data definition. This includes ensuring consistency between the data shared between modeling, testing, and reporting. Doing so will help you avoid implementation delays.

Where Is Operational Analytics Used?

The addition of operational analytics can have a positive impact on a wide range of industries.


Often, organizations are more fascinated with identifying new data sources to collect more data and lose sight of what’s in front of them. Their inability to make the most of their existing data prevents them from looking into potential opportunities in their sales tools, such as Intercom, Salesforce, and HubSpot.

Operational analytics can help them gain a better understanding of their existing data by creating more data flows within operational systems. The arrival of more contextual data provides more information about leads, allowing sales representatives to improve lead scoring. A better lead scoring program allows the sales department to target the right prospects with more accuracy, which can improve conversions.

Similarly, real-time and polished data can help salespeople segment their customers into different categories. This way, they can use the relevant messaging for their target audience — one that deals with their pain points.

Industrial production

Operational analytics can introduce predictive maintenance of machines and machine components. This can help detect issues before it’s too late. Here’s how this works:

  1. You choose a machine that often runs into issues and affects your production output.
  2. You review the machine history and failure modes to know why it breaks down (e.g., when motors overheat).
  3. You build an analytics model that predicts failure probability.
  4. You feed data from sensors that provide your machine’s data points for temperature, vibration, and other metrics to your model.

Over time, your model will use your machine history and the latest data to produce a relatively accurate estimate of when your machine is expected to fail. These models can offer two benefits. First, it helps you plan your maintenance in advance. Second, it can let you know which spare part you are likely to need in the near future and keep it in your inventory, speeding up the maintenance process.

Supply chain industry

Operational analytics can help businesses collect insights from the massive amounts of data linked to the procurement, processing, and distribution of goods.

For instance, consider a point of sale (PoS) terminal that uses a demand signal repository — a type of data warehouse or database that manufacturers use to collect PoS and other demand data. Implementing operational analytics can help you set up an ETL with the terminal to send real-time data to your repository and anticipate consumer demand with greater clarity.

By using prescriptive analytics in operational analytics, manufacturers can determine whether their partners are slowing them down. For instance, it can provide the history of a supplier from another country that has been delivering late orders for some time. Other factors can include diminished capacity and an unstable economic climate. The manufacturer might consider having a meeting with the supplier’s management to see if they can fix these issues or if it’s time to partner with another supplier.

What Tools Are Available for Operational Analytics?

A common issue with operational data is when different tools are unable to sync with each other. You have to ensure that your data flows in a reliable manner between applications. Some software platforms can move your data into a data warehouse, but transferring it from the warehouse itself is a dilemma. Data latency is another challenge. Even if you find a tool that can integrate with all of your tools, collecting up-to-date data in your dashboards is a struggle. That’s where Striim can ease your concerns by offering a plethora of capabilities.

Striim has real-time integrations for nearly all types of data pipelines. Whether you have to move data from a data warehouse or into one, Striim’s third-party support can ensure that your operational systems receive data more reliably.

Striim also lets you build customized dashboards for operational analytics. For example, one of Striim’s customers is an airline company that uses Striim to improve their daily operations.

striim for operational analytics
A real-world example of an architecture that supports operational analytics. Here an airline uses Striim to ingest operational data from various Oracle databases to populate their enterprise data lake and enable real-time decision making.

Striim’s real-time data ingestion ensures that the company can get their data immediately from Oracle databases. Next, Striim performs in-memory transformations on it and minimizes data latency by using their SQL-based stream processing capabilities. Next, Striim sends this data to Apache Kafka within milliseconds, and then the airline moves the data into their data warehouse, HDFS, and NoSQL database. Thanks to Striim’s role, the airline is able to use their KPI dashboards with data that’s less than two minutes old. Therefore, their management can make the right decisions for mission-critical flight operations throughout the world.

One of the issues the airline faces is managing their maintenance, especially when it comes to the de-icing process. In some cities, de-icing can take longer than expected due to snow incidents, causing flight delays.

With Striim’s dashboards, they can view real-time updates on how the de-icing process is going. This way, their team can adopt a proactive approach for managing potential flight departure delays. The dashboard allows them to make an instant decision, like whether to use a different piece of equipment to accelerate the icing process, assign another aircraft, or reassign passengers to other gates and flights. When making such decisions, it’s extremely important to get up-to-date, reliable data in the dashboard. Even a 10-minute delay in ingesting data to their dashboard can negatively affect decision-making. Striim ensures that they get this data in real time. Learn more about this case study here.

Get Buy-in from Your Employees Before Rolling Out Operational Analytics

Organizational change is one of the challenges organizations face when adopting operational analytics. The decisions made through operational analytics can influence manual decisions, which can be worrisome for the people who originally made those decisions. The insights from operational analytics can also empower junior professionals to make their own decisions. These employees previously relied on senior staff, so internal dynamics are bound to change.

Before implementing operational analytics, organizations should take employees into their confidence. Employees need to know how analytics can enhance their daily work by using automation to allow them to concentrate on the tasks that really matter.