Business Intelligence (BI) is meant to give insights for business users to help them make the data-driven decision making. The primary critical measurement of BI maturity is how much time is spent until an actionable insight is derived from data.
Traditional BI was started by separating the load on transactional systems in the form of data warehouses and giving access to the business data to users, mostly in excel reports. Even though it served the purpose of getting the information into the hands of the users, traditional BI applications demanded more manual time to analyze the data to generate actionable insights. For example, you can get answers to questions such as what happened, but to get answers to why it is happening, the analysts needed to spend more manual time. Also, traditional BI practices resulted in a lot of duplicated efforts in multiple business units, with each business unit having separate analysis practices with multiple excel file versions of the same data.
Self-Service BI tools evolved to address many of the traditional BI limitations with better visualizations, highly interactive, and connectivity to multiple data sources. In the form of centralized dashboards, all users can see the same KPIs and can consume the information quickly to derive insights. The ability to interact with visualized data helped users to not only get answers to common questions like what is happening and also why it is happening. Self-service BI tools tremendously reduced manual efforts of analyzation to derive insights.
BI Challenges But there are still gaps in most of the current practices. Let us look at them by exploring sales function with four levels of questions.
Level 1 – Business leaders want to know what’s happening with the sales? These are simple questions that can be answered by looking at a Sales Dashboard, and leaders can get to the numbers by themselves.
Level 2 – Leaders want to know why sales targets are not meeting in a particular business unit or for a specific product category? Here leaders are relying on analysts to get the desired information, and analysts can interact with data visualizations to find the insights causing sales decline.
Level 3 – Next, leaders like to know what will happen in terms of forecasts with sales and customer demands. Traditionally, leaders have to rely on exponential moving averages in excel reports. Now, they can ask data scientists to build a time series forecasting models with improved accuracy.
Level 4 – After looking at the forecasted numbers, now leaders have other questions like what the options are to achieve the desired target sales quota. In regular practice, leaders have to use their experience and brainstorm with multiple team members to come up with options, and then the final decisions are taken to act.
The important thing to notice from the above scenario is why leaders have to ask four different levels of questions and why do they have to work with various teams to get the desired actionable insights.
If we have a magic mirror, then we would ask, “Mirror mirror show me what I need to know?” or Show me what the most critical risks for my business are? Show me the best options for me to act? Well, with AI-enabled BI applications, we are almost there.
Before we talk more about redefining BI, let us have a quick overview of Artificial Intelligence. AI is an umbrella term for making machines cognitive like humans, and at the same time, more intelligent than humans.
The main branches of AI are the ability to predict from structured data, understanding unstructured data, and recognizing images. Machine Learning (ML) is the central part of AI, and in recent years ML gained much-needed attention and popularity. NLP is another branch of AI with the ability to process text (unstructured data) like in Google language translator and voice interactive personal assistants such as Alexa and Google Home.
Image recognition is another branch which is the crucial factor behind autonomous driving. Other branches of AI are Robotics to move physical things and also Robotic Process Automation (RPA) to automate rule-based process workflows. To keep the context-focused on BI and data-driven decision making, let us consider Robotics, RPA, and Image processing are out of scope for this article.
With inbuilt AI / ML features, BI platforms are maturing towards providing required insights by analyzing given datasets, enabling textual search for information from data, highlighting anomalies in data without you even asking for it. It is tempting to call these platforms as Intelligent BI, but I think Smart BI suits better.
Microsoft Power BI has a “quick insights” feature which runs multiple ML algorithms and presents insights visually. Similar to Power BI, Tableau has an “explain data” feature that runs various statistical algorithms and generates visualization pointing patterns, anomalies, and relations. Thought Spot is another BI platform vendor taking analytics to the next level by Search and AI-Driven analytics. Outlier.ai is another vendor claiming to jumpstart your day by delivering insights to your inbox without you even asking for it.
With NLP, most of the leading BI platforms provide the ability to search for data with natural language text form like “what are the sales from the south region in 2019”. Additionally, the platforms have the capability to give insights in a humanfriendly textual format instead of just showing visualizations with numbers.
Smart BI platforms can provide insights by pointing outliers and anomalies and trend changes and, in turn, getting you closer to the answers and reducing manual analysis for data-driven decision making.
Smart BI Competency
I am not just talking about the BI platforms’ capabilities but want to point out the direction of BI maturity and why leaders should prepare their BI competencies with the inclusion of AI/ML to address the business needs beyond descriptive analytics.
To go back to sales scenario, how about having BI dashboards showing the current state, timeline comparisons, trends, patterns, and also ML-enabled sales forecasting, clustering, risk scores, anomalies, outliers, and indexes. To provide that kind of dashboard, not all the calculations can be done within the BI tools, but dashboards can make calls to ML models on demand to retrieve required insights. Almost all the Smart BI platforms can call external ML models that might have deployed in the cloud or on-premise.
With that kind of visibility to insights, now business leaders don’t have to spend their time on basic questions but can focus on real questions such as what the best course of action is to exceed the sales targets. This approach also helps to improve an organization’s data literacy because users will start seeing a new level of data insights with predictive analytics and, in turn, will enhance organization’s capability towards more data-driven.
Rapid technological progress is happening in the area of Data, Analytics, AI, and ML. Smart BI platforms are revolutionizing the way to interact data for insights and see actionable insights.
Leaders need to start looking at BI differently. BI competency centers need to have team members with data science skillset to make the best of given tools and provide wholesome solutions to the business needs. BI and AI/ML don’t have to be two different competencies in an organization when it comes to analytics. AI-enabled Smart BI competencies will help companies to become not only more data-driven and also to become intelligent enterprises.