Embedded analytics in the age of the data-driven business
May 7 2019
by Krishna Roy
Embedded analytics, involving the use of reporting and analytical functions inside a transactional business application, has been around for decades. Business intelligence (BI) vendors have been offering reporting and analysis features that don't force users to switch to a different system to use them since the early 1990s.
But embedded analytics is now available in fresh forms, which weren't available in its early days. Furthermore, the wholesale corporate shift to becoming data-driven by making more decisions based on data (alongside a number of other factors, which we discuss below) means it could exhibit significant growth in the future. However, not every embedded analysis offering is equal, which is why we also set out a few criteria worth taking into account when considering this approach.
The 451 Take
Embedded analytics has long been held up as a way to democratize analysis by making it readily consumable by individuals who aren't analytically savvy. In the age of the data-driven business, that objective is even more critical. Market developments suggest this Holy Grail is moving closer to a reality. But the plethora of embedded analysis options now available call for vigilance when making purchasing decisions. Furthermore, the four criteria we outline below are the minimum to mull over. Security, scalability and vendor support are also worth delving into, while localization capabilities will be important for multinationals.
The growing popularity of data science for data-driven decision-making is a major new factor shaping the adoption of embedded analytics. One of the main challenges in data science is operationalizing the findings data science provides by making them comprehensible to individuals that need to act on the findings – or at least understand them. Embedded analysis is one approach to accomplish that because it enables outputs from data science models to be distributed in reports and visualizations.
Moreover, these reports, charts and graphs can now be folded into applications and tools that business users employ on a regular basis, which makes them easier to consume since the users don't have to swap apps, or learn a new tool, to see and understand them. The insights can also now be predictive or prescriptive in nature, because they are generated from data science models, thereby enabling more advanced forms of analysis to be more easily digested by those without data science smarts.
Embedding BI-oriented types of analysis into everyday applications that business users interface with is another growth area. It is predicated on the notion that if you are using sales, marketing, operations, finance and other types of business apps on a daily basis, that's the best place to deliver analysis as well. The benefit is faster insight to action because personnel in business roles need analytics in the context of their jobs to make truly effective decisions. Furthermore, this belief has been a driver for some significant deal-making activity in recent years.
Fresh forms of technology are also playing a big part in making embedded analytics easier, opening the door to this form of analysis becoming more ubiquitous in future. Application architectures are moving away from being monolithic. Decoupled, lightweight services or micro-services are now becoming the 'new norm.' BI platforms are moving in the same direction as well. Small, specialized BI services that can be purchased and deployed a la carte are becoming commonplace, paving the way for more flexible, faster and easier embedded analytic deployments.
IoT is also contributing to the growth in embedded analysis, driven by analytical functions integrated into devices. Manufacturers are using embedded analytics for predictive maintenance and other cost- and time-saving measures. Healthcare companies are employing this approach for diagnostic purposes.
The role of M&As in making embedded analysis possible is worth highlighting, so we have outlined a few early as well as recent deals – this list is not meant to be exhaustive.
Help desk provider Zendesk's pickup of BIME Analytics in 2015 is an early example of a front-office application vendor snagging analytics in order to offer it as an integrated part of its applications.
Workday is another business application vendor that has used an acquisition strategy to offer integrated analytics inside its human capital management and financial management apps. Workday nabbed Platfora in 2016 to kick-start this strategy. The company subsequently reached for Adaptive Insights in 2018 to bolster its Power of One strategy, which embraces embedded analytics, because it is all about providing one platform where all data is accessible – so individuals don't have to spend time on integration, and can plan, execute and analyze in one system.
Altair's acquisition of Datawatch in late 2018 had a significant embedded analytics aspect to it. Altair is looking to address predictive manufacturing and simulation requirements, and bolster its IoT analytics play using Datawatch's data science, analytics and integration assets.
Finally, Marlin Equity Partner-owned Logi Analytics' pickup of Jinfonet earlier in 2019 is worth pointing out. It involved one of the best known embedded analytics pure plays reaching for an early market entrant in this sector (focused on reporting), in order to consolidate Logi's market position as a purveyor of embedded reporting, analysis and predictive analytics.
Logi has retained its mantle as a go-to brand for embedded analytics, even though the company competes against a large group of BI and analytics vendors with offerings also pitched at use cases that are similar to its own. Indeed, most BI vendors tout an ability to integrate their analytical offerings into third-party environments – but the ease, speed, flexibility and effectiveness of their approaches can vary greatly.
Here are a few key things to look out for when evaluating embedded analytics offerings.
It may sound obvious, but it's important to ascertain the requirements of the audience you are deploying the embedded analytics to before selecting a product. It needs to meet users' criteria, otherwise it will be underused and ineffective. Do they need reporting and visualizations, or do they need recommendations (which calls for a machine-learning-driven approach that will be more complex to implement and maintain)?
Ease of connectivity and integration with a wide array of data sources is crucial. Successful embedded analytics relies on ensuring that all the data necessary for the analysis is readily available, otherwise the insights are likely to be incomplete.
An easily customizable user experience is vital. Most organizations adopt embedded analytics because it provides a transparent user experience to their employees, which is easily consumable because it doesn't require additional training or effort.
Potential purchasers should also evaluate the overall ease with which the embedded analysis can be achieved and maintained – otherwise development, management and administration headaches can ensue.