Linking analytics to key enterprise decisions
Enterprise analytics applications need to move beyond the analysis of an enterprise's past and present data; towards predicting the future behaviour of the enterprise, said Bradley Smith, founder and director BusinessOptics.
Speaking at the ITWeb BI Summit 2014 in Johannesburg, he noted that this should be done under different scenarios and recommending courses of action in these scenarios. These actions then need to be embedded into business processes to automate decision making.
To be able to do this, Smith expressed his view that the enterprise's processes and culture are key factors to consider.
"There should be in place a process that empowers business domain experts to create solutions that perform analysis, model future scenarios and recommend courses of action. Once domain experts are given reign to build prescriptive analytics solutions, enterprises will be able to tackle both new and existing problems from different perspectives."
Smith further pointed out that prescriptive analytics solutions empower business domain experts to build solutions themselves, which reduces their dependence on developers and increases the speed of exploration, insights, solution design, creation and the iteration thereof.
It can be extremely difficult to rigorously create upfront designs for analytical solutions to business, according to Smith, attributing this to uncertainty because of a lack of exploration of data.
"In addition, business domain experts are not programmers and do not know how to code, therefore they need to be empowered with a toolset that is visual, intuitive and doesn't require them to code," he says. "This toolset should be transparent, auditable and allow collaboration between other domain experts and customers."
To illustrate the integration of prescriptive analytics solutions into enterprise systems to enable decision automation, Smith used a sales and customer analytics example of a dynamic sales agent dealing in product recommendation and pricing for sales of non-trivial products such as insurance contracts.
"Logistics in this scenario include real-time analysis, scheduling and route optimisation (looking at operations from a global perspective) with instructions being sent electronically to dispatch units and vehicle operators," said Smith. "Now, the challenge here is that selling to people costs money and not all customers are profitable, and enterprises want to select the leads that they expect to make the most profit given the marketing budget."
The solution, he stated, is using predictive analytics to learn from demographic, financial and product characteristics to forecast the customer lifetime value (and thereby profit) from a particular customer. He suggested companies then contact leads that are expected to enhance the overall profit from the sales campaign.
"With this information at their disposal, the enterprise can offer products, features and discounts that are expected to maximise the chance of sale and profit."
He concluded that embedding prescriptive analytics within enterprises for real-time decision support and automation can provide significant benefits with respect to better decision making, streamlined processes, lower costs and ultimately, improved revenue.