Technology continues to rapidly transform every industry. The world of BI and analytics is no different. We’ve shared five predictions every data leader needs to know about in 2019, but there were five more we couldn’t leave out.
6. Semantic models make a resurgence
Traditional big business intelligence products introduced the notion of semantic models - data models that aimed to provide a centralized, unified, logical, business-oriented abstraction over physical data sources, which serve as a single source of truth for business metrics. While the motivation for using semantic models was sound, the tools for building and maintaining them were extremely complex, requiring very specialized and expensive resources. These models were managed by a small team, creating a serious bottleneck to change. Even more cumbersome, these models were proprietary, meaning each tool had its own model. Enterprises often had multiple such tools, resulting in multiple “single-sources-of-truth”.
Second generation enterprise business intelligence tools bypassed the failed centralized semantic model; however, they ended up creating decentralized models per report/analysis. As these tools proliferated within various pockets of enterprises, they ended up creating an even bigger mess of scattered, isolated, ungoverned, insecure, localized models.
There is huge value in a centralized, single source of truth. This is the only way enterprises can be efficient, accurate and effective in their analytics. Centralized semantic models are a good thing. But they need to be built differently in 2019 and beyond to be maximally effective. They need to be:
Focused purely on the concern of providing a common, logical business view over multiple physical data entities
Able to span on-premise, private cloud, and public cloud sources
BI tool/vendor agnostic, so any tool can work with it using standard interfaces
Capable of being automatically generated by software that can introspect physical sources
Built collaboratively through crowdsourcing with robust version tracking and audit workflows.
7. Data teams create less pre-built reports and dashboards
With self-service reporting capabilities becoming a reality for more and more business users, the need for canned reports and dashboards, produced by centralized data teams, will reduce significantly.
Self-service BI breaks the classic producer-consumer model of BI. Business users no longer need to depend on central teams to produce reports and dashboards for them. They can get the insights they want, whenever they need, on their own, with search. What’s more, business users will end up creating more BI reports and dashboards for themselves. These reports and dashboards will be personalized for individuals or groups, easily adaptable, and highly relevant for users making day-to-day business decisions.
8. Data Science is democratized for the masses
Data science, including AI and ML, has been the domain of a few highly specialized, technical groups within enterprises. While there is no doubt that data science has added huge value for companies, this value can be multiplied several-fold with the democratization of data science. Democratization is achieved by making the power of advanced analytics available to business users who likely are not experts in data science.
For example, a data science team may develop advanced customer segmentation algorithms that are exposed as formulas or search terms for business users to use in their analyses. The user is unaware of, and unconcerned about, the underlying data science algorithms that powers the terms that she uses.
Democratization will not diminish the role of data scientists. On the contrary, it will increase the demands for more advanced data science algorithms, as more and more business users leverage the power of data science to find powerful insights.
9. Standalone data integration tools gain prominence
At the core of any analytics product is data. This data is likely to originate in many different sources and needs to be rationalized before becoming useful for analytics. Data integration is the process of periodic data movement from transactional systems to analytic systems, cleansing, enriching, de-normalizing and linking them. Data integration is a specialized area, with many vendors providing sophisticated products and solutions. A new breed of self-service data integration tools, such as Alteryx, have gained prominence over the years by democratizing access to data integration for business users. These tools have also become very rich, employing advanced AI and ML techniques for everything from automated data cleansing to relationship detection.
BI vendors have attempted to build data integration tools within their product suite. While this may help small companies that want one single, simple solution for all their needs, it does not work well for large enterprises. Large enterprises have complex data integration needs and are likely to lean more towards specialized vendors that offer all the rich capabilities that they need. In 2019, we will see the demand for data integration capabilities within BI systems declining, while the demand for best-of-breed, standalone tools that can easily integrate with BI systems increases.
10. Social data platforms promote a more fact-driven world
As 2018 clearly showed, we live in the age of fake news and distrust of information. The consequences of this can be disastrous for governments, the environment, and the general progress of humankind.
How do we combat this? By cultivating a culture of fact-driven discussions on all important topics that impact us. A powerful way to cultivate this culture is to provide a free and open platform where people can analyze data easily and socialize insights with others. Others are free to do their own analyses and provide their points of view, but the same data is available for everyone to verify.
Data always speaks the truth. A business intelligence platform that helps everyone expose the truth widely, backed by solid data, can help promote a more fact-driven world. Such BI platforms will gain more prominence, not because of commercial interests, but because of a global realization of the negative consequences of not being fact-driven.
What are your predictions for the BI and analytics market in 2019? Sound off in the comments, and check out part one of this blog series for five more things to expect next year.
