Business Intelligence: Custom tools to a simple search box
Business Intelligence is simply getting actionable insights from existing data and taking decisions based on the same.
What would the ideal design of a BI system look like?
Consider the calculator. The human part involves entering the input and the operators(+,-,*..) to the tool. The calculator processes the data and returns what is expected. It is a perfect complement of tools and human beings applied in a manner so that the whole is greater than sum of its parts.
Going forward, BI systems are expected to evolve into a search based system that uses semantic search and Natural Language Processing. This would enable users to put questions either through a text box or a voice enabled input system.
That would ideally involve a person asking a question such as ‘What is the revenue and growth for the past 5 years?’. The answer would be a graph, diagram, table or any other format that best helps a person being understand and visualize the revenue and growth for the past 5 years. This would then help in making decisions at a strategic or operational level.
How is BI implemented currently in the industry?
I currently work in an IT services firm providing business intelligence and analytics solutions to clients across multiple industries globally. Almost all our solutions involve building reports or dashboards that take data in a regular fashion and display results in a pre built manner.
The solutions we provide are for two sets of users. Either they are from business users who are looking for reports or dashboards that are consumed on a regular basis. If they want a new kind of report or include additional data later, they need to reach out to IT for newer reports or dashboards.
The other category involves power users who use tools meant for self service BI. This involves charts, dashboards and other tools to quickly help you create meaninful visualizations out of your data. QlikView, Tableau and Power BI are examples of these.
Is text box input model used anywhere currently?
One tool where we have seen is in Power BI for desktop, where you can ask a question such as ‘What is the sales and costs for June 2016?’. If there are column names that match to ‘sales’, ‘costs’, and ‘June 2016’ in the data that has been entered to Power BI, the results will show. You are not going to get an answer for ‘Give me the profitability for last month’. But you will get an answer to ‘What is profit/cost percentage?’.
In the ideal scenario, you are looking for the certain number from a set of data that has been made available to a tool. The tool would process your inputs and return a certain output on which you can take decisions.
Going from current scenario to search based BI
The adoption rate is slow with fewer users going for a self service environment. But the fundamental needs when approaching a BI solution remains same and will possibly continue to do so going forward. They are analytical capability for deeper insights, visualization for easier understanding, integration or connectivity with multiple data sources and support.
For a BI tool, what do most customers look into?
Analytical capability: Analytical requirement would vary across industries. For example, for a retail client, this might look like a single customer view, channel sales, RFM analysis, life time value, market basket, insights on previous interactions etc.
Visualization and easy-to-understand reports: Easy visuals and more charts are always the requirement while choosing a tool. For the kind of information being presented, it should also be visualized in the right manner to help derive value and take actionable insights.
Integrations (Data connectivity): Usually, clients have existing data sources and are looking for connectivity for the same so that existing systems don’t have to be deprecated when new systems come in.
Support: Support from the vendor going forward and capability from the IT services provider is looked into by most clients before choosing a tool or a services provider.
The rate of adoption of newer BI systems in the industry remains slow, but the fundamental requirement continue to remain the same and will do so going forward.