Embedded Analytics Evaluation Guide

This paper will cover user expectations of modern applications and how they have amplified and heightened over time, including key drivers to staying competitive.

It will take an outside-in approach to understanding the critical steps to embedding analytics and the central considerations for deploying scalable and successful embedded analytics to meet elevated client expectations of modern applications.

Drive additional revenue and stay competitive

Organizations across the world are recognizing the power of data to not only drive efficiencies but to also drive additional revenue and growth. 

With the increasing demand to be competitive, businesses have turned to the advancements in analytics to capture new opportunities and markets. Now customers have come to expect and require an analytics component as part of the product or service offering. Whether it’s providing analytics on data that is already collected or creating a completely new product, businesses have the opportunity to delight their current and future customers with fewer financial resources than ever before.

Based on the business model, the pressing need might differ. As a software company, the imperative could be to go on the defense and build a front-running competitive advantage with innovative insights. As a hardware company, the imperative could be to go on the offense and help increase margins and drive new revenue streams otherwise left uncaptured. Or as a services company, the imperative might be to scale and move away from low-return custom solutions.

Whatever the reason for embedding analytics, not leveraging data is tantamount to throwing revenue away or losing to the competition. 

Embed analytics to win

Today, it is common to embed an analytics solution by working with a trusted partner and embedding its product into one’s stack. Embedding analytics is proven to be the easiest and most effective way to offer analytics to your customers, allowing you to bring a quality product to market faster and with fewer internal resources.

Leveraging an analytics provider to embed analytics into your product allows product and development teams to focus and innovate within their core business, rather than spreading resources to build products outside of their expertise. Additionally, resource allocation and budget generally surround monetization of embedding analytics, such as how to increase sales, reduce churn, or provide new revenue streams. 

Data to revenue and retention

Data to revenue and retention

The steps from data to revenue and beyond include more than just selecting the technology and platform. All three pieces — people, process, and technology — need to come together to make that journey. 

This journey includes identifying data opportunities, building the P&L and use cases, getting the right buy-in, finding a partner, implementing and launching, and supporting and growing the business through an iterative process.

As with any product, innovation and growth are critical to stay ahead of the game. 

Assuming you have discovered data opportunities and are looking to drive revenue or efficiency with embedded analytics, you must then ask yourself what the important considerations are to ensure you are building seamless and scalable analytic applications.

Heightened expectations from modern applications

Heightened expectations from modern applications

Today’s users are accustomed and expect seamless and customized experiences that are fast, easy to use, and add value. Enterprise clients and end users of business applications are no different!

Meeting and exceeding these client expectations requires the right infrastructure, application architecture, and processes.

Delivering seamless experiences requires customization and extension capabilities, API-driven integrations, and the ability to deploy the application in any workflow.

A quick response to needs requires a powerful and unified data engine along with an agile and easy-to-use platform for both the developer and the everyday user.

The solution is to build applications using modern processes of scaling, which are DevOps enabled. 

Staying competitive in the modern application world

Staying competitive in the modern application world

To stay competitive in today’s world, applications should be able to leverage and support these critical shifts:

  • API-driven workflows and ecosystems: Developer-first platforms are critical in competing in the modern world. You have to be able to integrate, customize, automate, and extend to provide the best possible business outcome for your end users.
  • The age of the cloud ecosystem: Highly scalable applications are built keeping the cloud ecosystem and DevOps in mind. This becomes even more critical when you are delivering apps to customers, who will have less patience for a bad user experience than internal users.
  • The rise of analytic applications: AI- and machine-learning-powered analytics have to be an integral part of applications. The line between operations and analytics will blur. Analytic apps will provide embedded actionable insights with highly interactive, visually pleasing, and rich experiences.

Shifts in application development paradigms

Staying competitive in the modern application world

These shifts in application development could not happen without the support of what’s beneath the hood.

To support the scale and seamlessness that we are used to and have come to expect, modern applications are typically built using new paradigms — whether it is building a cloud-native architecture with an API-first approach, or leveraging the power of highly scalable data warehouses and data storage.

You might be at the beginning of this journey, or you might be on a born cloud-native platform. Wherever you are in the journey, it is important to consider these factors as you look into embedded analytics for your product or application. 

Embedded self-service analytics

When working with an embedded analytics application, a user experiences the interface. But as with any application, what the user sees and interacts with is just the tip of the iceberg.

Seamless experience

To provide the seamless experiences that your customers expect, a lot has to happen below the surface — automation, user security and integrations, scaling, self-healing, governance rules, and more. 

Building standard visualizations is a commodity today, and so is dropping them in an application.

To be truly successful, there is more to it than just creating a few visualizations in an application.

Key steps to evaluating scalable embedded analytics solutions

When implementing embedded analytics, there are broadly five categories of steps (which can be performed in parallel as well):

  • Deployment model: Designing and setting the deployment model and architecture
  • Data and self-service insights: Building the data models and visualizations with customizations if needed to meet unique needs, or closing the BI loop by adding action workflows to third-party applications and systems plus providing your customer with the ability to do more on their own in the context of their application, whether through self-service or self-design capabilities or AI-assisted workflows 
  • White-labeling and embedding: Matching your required branding needs and embedding insights into your product or application
  • Security and integrations: Ensuring that customers and tenants only see what they should and ensuring seamless security sign-ons
  • Scalability and high availability: Making sure your application is primed for scale and high availability to ensure continuity and performance

Underlying all this requires the right foundation of a flexible and open platform.

In addition, automation has to support every step, and you need the right partnership and support to ensure a smooth and successful process. 

Let’s break each down and understand some important considerations with each.

Deployment model

Deployment model

The question to ask here is if you are set up for change and growth

  • What does your application and data store deployment look like? Is it single-tenant, multi-tenant, a hybrid, or in transition?
  • How quickly do you need to be up and running? What are your growth plans?
  • Are you transitioning to the cloud? Do you plan to or see a future there?
  • Do you see the need to prototype and innovate with new data?
  • Do you see customers wanting advanced access to data directly?
Deployment model for Sisense data analytics

Change is the only constant, and in the fast-moving world of business requirements and applications, you need a platform that can meet what is thrown at it.

When looking at flexibility of the platform, traditional approaches typically tend to support a narrow set of options.

Say you have the need for:

  • Operational analysis to be worked off cloud data warehouses as well as historical data analysis where speed is more important than freshness of data
  • An environment where customers can add their own data or an environment for innovation or rapid prototyping without building data pipelines first
  • A mix of deployment styles, with some of your clients on separate installs and some on a multi-tenant deployment

It is critical to have the flexibility to achieve what you need in these scenarios. 

Product teams must think about flexibility on all levels:

  • Data: Can you connect to any data, your way, at scale?
  • Deployment strategy: Can you deploy on any cloud or on-premises or a hybrid setup in a single-tenant, multi-tenant, or hybrid environment based on your application deployment?
  • Infrastructure: Can you deploy on any cloud or on-premises or a hybrid setup with the ability to move in the future?
  • Content delivery: Can you deliver context on the web, in applications, and via mobile, with various levels of embedded analytics sophistication?

Traditional architectures tend to be older-school and monolithic rather than truly cloud agnostic and built for scale. This means they can’t truly leverage the benefits of cloud architectures and processes and move to the future with ease.

  • Can you move the architecture to another infrastructure without any hassle?
  • Is the architecture built on open standards and fully self-contained? 
  • Do you need special tools and software to build your solution?

Data and insights

Data insights

Building standard dashboards with bar charts and line charts is commoditized today. The real question is, how do you take it beyond a standard experience and delight your customers with innovative, seamless, and valuable experiences? 

Customer delight comes from enhanced experiences that provide value at the right time. There are three components to this:

  1. Rapid response to client needs: Are you able to innovate quickly and meet various client needs?
  2. Innovative and cutting-edge solutions: Are you going beyond traditional historic reporting? 
  3. Self-service: Are you supporting self-discovery and the ability of your customers to get deeper insights faster?

As you build your data products, ask yourself:

  • Do you see customers wanting self-service capabilities in the context of their applications?
  • Do you want advanced, predictive, and AI/machine-learning-based analytics for your customers?
  • Do you see unexpected and unique requirements coming up?
  • Will your analytics need to tie into other systems and applications to close the BI loop?
  • Do you want to reduce the burden on the development team to answer questions and save time and resources?
  • Do you want to enable your customers to build and design dashboards on their own?
Data and analytics evaluation questions

End users today want to get to deeper insights fast in the context of their workflows. Ideally, they do not want to go back to the embedded analytics provider for every new analytics question they have. Having their vendor as a bottleneck only adds friction and increases time to value.

Traditional platforms typically support static embedding without an underlying data engine to support dynamic data discovery. They are heavily dependent on data and tech teams to serve up insights to the customers. In addition, they do not provide AI/machine-learning-driven capabilities and advanced analytics to go beyond standard report generation.

Automated and augmented insights will enable your customer to get insights without data scientists and data teams. This will not only free up engineering resources to take on more difficult challenges but also deliver enhanced experiences to customers. 

It is also critical to be able to have the flexibility to support any given use case needed by the customer. While it is great to have a rich set of out-of-the-box options for data and analytics functionality, it is near impossible to account for every possible need and request. For example, there might be a particular domain-based need for a specific chart or an integration workflow back into a third-party system. The platform should be extensible and customizable to adapt and meet customer needs.

White-labeling and embedding

White-labeling and embedding

The best kind of analytics are invisible. Actionable insights should be delivered at the point of decision-making — whether that is in a customer relationship management application, an email, mobile notification, or on the screens at a bedside.  

As your embedded analytics matures, you should be able to blur the lines between analytics and your operational workflows further and further. For example, the simplest first step might be to white-label and deliver the analytics as-is. Next you might deliver an embedded dashboard in an iframe. A next step could be to kick off actions in the embedded dashboard from the host application. After that, you could embed individual analytics elements into your application and then tie the workflow back into your application’s workflow. A further step might be to embed pure natural language queries into your application. Sometimes, it can be a combination of all based on the experience you are trying to provide.

Embedded analytics should be an essential part of your application development and should have the frameworks and tools to be fully integrated into your platform and application. 

Embedded analytics - the old way and the modern way

Traditional applications usually provide very limited embedding capabilities built on just the iframe and will not have the ability or will have limited capabilities to white-label, extend, and customize the application including the mobile experience.

Your product team might have different steps to go-to-market based on your resources, needs, and progress in your strategy.

So for example, you can get to market with almost zero development time or without an embedded application by just white-labeling and changing the analytics application, or you could build a completely custom application experience with analytics blended into the experience. 

Another important consideration is the look and feel and branding requirements to ensure a seamless experience for your customer end user. The platform should enable updating the look and feel and white-labeling easily and quickly.

Ultimately, you need to be looking for a platform that is focused on embedding capabilities as a core of its product and provides you with various options to support your go-to-market strategy and maturity model within your resource constraints with the ability to do more as your data product scales.

Security and integrations

Security and integrations

When talking about embedded analytics solutions, security integrations cannot be far behind. It is critical to tie embedded analytics to your ecosystem to not only secure the application and data but also provide a seamless user experience. 

It is important to ask yourself how you are going to provide that experience. What tools do you have at your disposal to make sure that your customer and end user doesn’t have to log in again while not worrying about providing a secure experience?

In addition, your product will grow in scale and complexity, so how can you support that scale with minimal risks and effort? Manual workflows are impossible.

Security for embedded analytics - the old way and the modern way

Security is a criteria on many levels, but let’s focus on two issues that are most pressing for an embedded analytics implementation — user security and roles, and data security.

Embedded analytics require fine-grained security at multiple levels of the application from the system level to analytics level to row level. For example, can you ensure that one group of users gets to see a slice of your data model and only that? If you are dealing with a multi-tenant dataset, how do you split the multi-tenant data to enable reuse? You want to ensure that users, groups of users, or tenants only see what they should see. This is even more critical with embedded analytics due to multiple tenants who are different customers. There are two types of security — access rights and data-level security. Access rights have to exist at multiple levels, from the visualization layer to data layer to the physical layer. Data security splits datasets down the row level.

Secondly, you need a simple and clear API architecture along with extensive and modern single sign-on capabilities that will enable automation of the integration workflows. 

A unified governance layer is also critical to ensure a smooth process. Several platforms have silos of workbooks or data that sits in multiple places, creating governance nightmares.

As a member of the product team, you need a platform that provides API-driven workflows across the various levels of the application to ensure that end users are provided access and can only see what they are meant to.  

Scalability and high availability


Product teams are most likely concerned about four other important factors — a large volume of concurrent users, big data, uptime, and disaster recovery — all while maintaining high performance. You need a solution that can grow with you as you gain more customers and users.

An often-overlooked aspect to delivering embedded analytic solutions is how they will be future-proof. Unlike an internal use case, your number of end users is not limited. In fact, if the product does well, you want to be able to get better with it over time. 

Scalability for embedded analytics - the old and the new way

There are three important considerations when talking about scale and high-availability.

1. Performance and availability

Traditional platforms simply are not architected to support dynamic allocation of resources, nor are they purpose built for large amounts of data that doesn’t require a ton of IT or hardware resources. Modern platforms are architected to increase resources on-demand, and self-heal with the ability to analyze data directly on data warehouses for scale with the option of cached data for minimized query costs, flexibility, and rapid prototyping.

2. Automation — a critical component of management and scale 

Adding users, ensuring security, duplicating and managing data models, sharing dashboards, installing, sharing, and moving assets from one environment to another are critical components of deploying an embedded analytics solution. For true scale and efficiency, all of these processes should happen automatically.

3. Monitoring and system metrics observations

This involves the ability to measure and detect system resources consumption and avoid outages to ensure operational continuity.

Support and partnership

Partnering for embedded analytics - the old and the new way

An oft-overlooked component of taking embedded analytics products to market is the importance of partnership. The relationship with your vendor cannot be a one-time transaction. Partnering with a vendor that has the expertise and experience in this space with a focus on customer success is critical in ensuring you are set up for success.

Look for:

  • History of embedding analytics customers
  • Customer success, industry, and analyst ratings
  • Resources and programs available to jump-start and support your growth phase in embedded analytics and data monetization

Business levers

Business levers for embedded analytics - the old and the new way

Time is money, and it could never be truer in software development projects. It is important to consider how long it will take to bring analytics to the market not just for the first time but with subsequent updates as well. 

  • How much do you want engineers to be focused on delivering analytics versus other core development projects?
  • How do you enable constant updates and changes without going back and editing code?
  • How do you enable your customers to do more on their own so that time to value is minimized?

Ultimately, as you bring a product to market, you need to think of the monetization methods and steps. What are the various ways in which you can deliver analytics to your customers? How do you build a tiered approach to what you offer? Will the platform support you in the various options you intend to build?

Key to successful embedded solutions

Keys to successful embedded analytics

To summarize, to successfully deliver, maintain, and grow your embedded analytics solution, you need a platform that supports all four personas. Often, a platform offers one or the other but not all of them. For diving deeper while still scaling, you need to be able to support the tech and the business side.  

  • Developers: The platform needs to be open and flexible to enable developers to integrate, automate, scale, and extend the platform as they need without losing full control.
  • Data analyst: For the more complicated questions and custom deep dives into the data, data analysts and engineers should have the ability to use code-first approaches.
  • Designer: The platform needs to support the everyday user (in many cases, the customer themselves) through ease of use, self-service, and data democratization, rather than requiring the engineering team to build the analytics layer. If the engineering team is taking on the mantle, this process should be fast since they shouldn’t be wasting their time reinventing the wheel. 
  • End user or consumer: The consumer should be able to do more than just look at a chart. They should be able to self-service, filter, slice and dice, drill down, ask further questions through natural language, perform actions through a closed BI loop, and get advanced predictive insights to go beyond traditional analytics.
Learn More