6 key features of embedded analytics software

Key embedded analytics software features

1. Pricing model

Organizations should consider what kind of pricing model works for them when choosing embedded analytics software. Embedded BI can decrease the initial costs of adding reporting, data visualization and analytics to data systems because an organization doesn't have to build the systems from scratch. BI vendors typically take care of tool updates, which reduces ongoing maintenance costs.

Embedded BI services delivered via cloud are even easier to implement than on-premises options because the organization doesn't need to install and run hardware. The organization incurs the cost of adding the embedded dashboards to enterprise software or writing code to call the BI vendor's application programming interfaces. Additionally, cloud services have subscription costs.

Embedded BI vendors typically calculate their costs using a combination of user seats, hosting and usage, which can add up quickly. To get an accurate estimate of costs, consider both typical usage and peak usage periods.

Another pricing option calculates the fee based on how many servers a company runs the tool on. In theory, a company could use a low number of servers to support an unlimited number of users. In practice, that can lead to poor performance during peak times or force the company to add servers.

Beyond pricing implications of cloud-based or SaaS delivery of embedded analytics, embedded tools can change or break when vendors upgrade their system.

"Once you get into SaaS for embedded analytics, it becomes a very finely choreographed process," said Boris Evelson, vice president and principal analyst at Forrester Research. "You just get the API, but you're at the mercy of the vendor to do the releases, do the security and do the upgrades on the same schedule that you're comfortable with."

2. User experience

Intuitive interfaces can reduce the time companies spend on training, increase embedded analytics usage and help the company become more data driven.

Personalization enables end users to choose the visualizations and reports they need. Many BI vendors make it simple to embed interactive dashboards into applications, allowing individual users to change their layouts or to drill down into the data. This is beneficial for organizations that want the ability to customize their platform.

For example, some customers of Stream Financial Technology want to use data from its APIs without using its interface, said Jowanza Joseph, former head of engineering. The customer decides how the data from Stream looks and how they embed it.

For some users, having mobile access to embedded BI platforms is important. Carefully evaluate vendor options because not all dashboards work well on mobile devices.

3. Predictive capabilities

In the past, BI platforms told users what was happening at their company. For example, BI can send a notification when the inventory levels of a particular product are low. Today, that's not enough.

A key feature of modern embedded BI platforms is the ability to make predictions and customize the tool to fit specific business needs. Stream Financial Technology didn't originally offer predictive capabilities. Customers wanted more finance management tools to predict future cash flow.

4. Actionable insights

In a standalone BI application, insights are valuable, but they're not really actionable.

It's relatively simple to embed a chart or dashboard from a BI platform into an employee application. The difficulty is making the chart interactive and actionable, Evelson said.

"In a standalone BI application, insights are valuable, but they're not really actionable," he said.

The BI platform can notify an employee that a particular item has low stock. The employee must open the inventory management system and order more inventory. Embedding BI into the inventory management system saves the employee time.

"If [the BI system is] well-designed, it will give you a list of the best suppliers and you can order the inventory right there," Evelson said.

Embedded BI users want to do more than predict future outcomes; they want analytics tools that offer "predictive capabilities that support proactive action and automation," Henschen said.

For example, instead of telling employees that inventory levels are low, embedded BI can tell employees that inventory levels will be too low to meet seasonal demand next month. The early warning encourages the team to put in the order right away.

Organizations can track the usefulness of actionable analytics depending on the outcome of the action. In a standalone BI system, the employee goes somewhere else to take action and it can be difficult to link how much the analytics helped.

Actionable analytics amounts to more than embedding a chart from a BI platform. The business application needs to make a call to the BI platform via an application programming interface, but not every BI system exposes all its functionality via APIs.

"In a modern BI platform -- one created natively for this purpose -- 100% of the endpoints are open to APIs," Evelson said. "But most enterprise BI platforms still have legacy baggage and only 70, 80 or 90 percent of the platform is open to APIs."

When choosing a BI platform, companies need to find the functionality they want to bring into their other apps via APIs. They should make sure the vendor they choose has that functionality or plans to add it.

5. Automated insights

According to Gartner, automated insights are the top functional requirement today for an analytics and BI platform.

Organizations can use machine learning capabilities to organize and understand the evergrowing data sets used by analytics and BI tools, said Colin Reid, a Gartner analyst, in the August 2023 report.

Embedded BI can automatically sift through data sets, generate reports, identify key information and create visualizations. Automation can identify insights and trends faster than humans could in large data sets, and it can reveal patterns that a human analyst might miss. It frees analysts to focus on more important tasks and makes it easier to distribute reports to stakeholders in a format that anyone can understand regardless of technical skill.

6. Integrate natural language processing and generative AI

Embedded analytics are constantly changing as GenAI and large language model developments improve capabilities to understand plain-English questions.

Natural language processing (NLP) has limitations in what it can do, said Bradley Shimmin, chief analyst for AI platforms, analytics and data management at Omdia.

For example, many BI systems have nuanced definitions for their data and insights. A measure and a metric might mean different things. As a result, trained data scientists must set up embedded BI and pull in the insights needed in each context. Adding NLP didn't help.

With the advent of GenAI, users don't need to know the difference between a measure and a metric or even the data models and assets their organization uses, Shimmin said.

"In the past, no matter how good the natural language processing was, you still had to know the nuances of how to phrase a question properly," he said. "Generative AI is rapidly becoming the realization of the work these vendors have been taking over the past five years with NLP to try to put a human experience in front of a very complicated interface."

Embedded BI tool growth

Software companies use embedded BI advancements to add analytics and BI capabilities to their existing platforms without building them from scratch. Vendors can get tools to market faster and focus on the core features of their software.

According to a report from Constellation Research, the top vendors in the embedded BI market include Domo, Google Looker, Microsoft Power BI, MicroStrategy, Oracle Analytics, Qlik Sense, Sisense, Tableau and ThoughtSpot.

Organizations such as banks and retailers can add analytics to their apps, portals, online loyalty programs and other software. Nontechnological organizations require more features from BI vendors as they move from simple charts to more complex and interactive BI features, such as making decisions based on analytics insights.

"Microservices architectures, fine-grained application programming interfaces, RESTful interfaces and software development kits support flexible embedding of data, metrics, visualizations and dashboards," Henschen said.

Commercial platforms can also support DevOps approaches that automate continuous integration, continuous deployment, low-code or no-code development and embedding options. Organizations can use DevOps capabilities to build custom applications and help nondevelopers add data and insights to existing applications. Enterprises buying embedded BI tools should also look for alerting and workflow options.

"These enable organizations to trigger events, automate actions and kick off workflows based on data- and insight-driven rules and thresholds," Henschen said.

Maria Korolov has been covering enterprise technology for nearly 20 years and is currently focusing on artificial intelligence and cybersecurity.