Embedded analytics brings ease and improves the relevancy of data which all help enterprise end-users execute work tasks more efficiently and timely, but this may come at a compromise in terms of information "myopia" because the scope of analysis rarely extends to other data sources outside the embedded analytics tool, say observers.
Surya Mukherjee, business intelligence analyst at Ovum, defined embedded analytics as a broad term referring to various approaches and technologies that allow business users to perform analytic operations without leaving an enterprise application environment. That is because the business intelligence (BI) is being operationalized, integrated and used as an indistinguishable part of the enterprise application, he explained.
Dao Thi Minh Thao, ICT research associate, Asia-Pacific at Frost & Sullivan, highlighted a difference between in-memory and embedded analytics. The former is a technology approach to data warehousing, where data resides in a computer's random access memory (RAM) instead of hard disks.
Embedded analytics, on the other hand, can be involved at any stage of IT--not just data storage--that will overall improve a business process, making it easier to sieve out and interpret valuable information to users, she said. This ultimately helps deliver better decision-making and timelier operations of an organization on a daily basis, she noted.
Benefits of ease, relevance
According to analysts, embedded analytics have been around for at least more than a decade but only in the last few years gained more prominence among enterprises, particularly given the growing volume of data.
Dao said companies have only just started to realize how embedded analytics can optimize their overall business performance. Apart from helping decision-making, embedded analytics also help decrease the amount of irrelevant data stored to further simplify data analyses.
Mukherjee explained that the fact that embedded analytics are an indistinguishable part of an application is beneficial to users. "Consider a CRM (customer relationship management) user who requires frequent reports on customer value, an ERP (enterprise resource planning) financials user who wants to test the effect of daily market changes on the company's stock, or a HR (human resource) executive who wants to be alerted before any staff creates conflicting flight schedules.
"None of these users would typically want to invest time and effort in learning a new BI system or interface. Embedding the analytics insider the respective applications they use every day while abstracting the plumbing away from them, they can go about analyzing the data relevant to them, without having to learn additional tools," he said.
Shyam Baddepudi, vice president of business solutions at SAP Southeast Asia (SEA), agreed, saying the embedded analytics should not require extra training for business users.
He pointed out that embedded analytics has always been in use in certain critical areas of business application which delivered high impact to business, and were used only by those who develop business strategy in an organization.
However, with strategy-to-execution cycle times shortening in today's business world, various users in an organization cannot wait for critical BI to be generated for them before they can take action. Often, by the time the insights and information are distributed to the respective workers who need it, the opportunities have slipped away, he said.
By nature of having the analytic information embedded, presented and consumed within specific, real-time context as users interact with an application, embedded analytics helps users of the corresponding business roles make faster decisions and take action, instead of waiting days or weeks to act on their jobs more effectively, he explained.
This makes embedded analytics particularly useful in environments filled with a "lot of action and dynamism", such as the trading desk or call center, said the SAP executive.
Issues of customization, data "myopia"
Despite the benefits, there are also challenges as well as limitations with embedded analytics, industry observers cautioned.
Mukherjee said consistency in terms of analyses could be problematic, if the data source for the embedded analytics is not the same as the other external BI applications in the first place. Apps with embedded analytics may have access to raw data that is current whereas BI apps will likely access a "cleaned" data warehouse that is updated less frequently, he added.
At the same time, since analytics is embedded within an app, if the result analysis does not extend to or involve other data sources, it may "suffer from information myopia" and reduce the "depth of the analysis", he added.
Dao pointed out another issue which is that embedded analytics requires a high level of customization, which means enterprises have to dedicate resources to define the business process clearly and have consistent IT spend.
SAP's Baddepudi said in order for an application user to see and understand data and immediately take action, there is first the challenge to connect transactions and analytics into a single composite analytic application--which should also be easy to use and not require additional training for users. He added that having an integrated platform to begin with helps in faster deployment.
This can also ensure that insights are provided without additional delay that could be due to fetching and analyzing information while the user is interacting with the system simultaneously, he noted.