By now you understand that when it comes to analytics use cases, mind alone tends to be too expensive or too slow. Machine alone can’t deliver real insights or knowledge. Success lies in knowing how to mix the two. However, the key lies in knowing how to find the efficient frontier.
In reality each use case has an optimum mix of minds and machines that will provide the most efficient way to achieve the goals and deliver the best return on investment. Over-automating a process can lead to unnecessary cost while no value is added to the use case. On the other hand, under-automating might not deliver the desired client benefits or might simply render the use case uneconomical.
Investment professionals call this the efficient frontier, i.e., the optimal mix of various assets that yields the best returns. The rules for applying the efficient frontier model to investment banking were developed by Harry Markowitz, who won the Nobel Prize for his theory in 1990, but they can be applied to analytics use cases too. Let me show you what I mean.
Rule 1: Analyze the end-to-end use case for automation potential. The most common mistake is to focus on the data and analytics, but forget the dissemination to and interaction with the end users and knowledge management. Think of every use case as getting oil from the well to the gas pump. It isn’t just the work in the refinery that counts.
To make sure you don’t miss anything, remember the five areas of automation that can be applied to each use case:
Data and analytics tools, including AI
Publishing and dissemination engines
Knowledge management tools
Rule 2: Keep it simple. At the beginning, focus on the low-hanging fruit and create a minimum viable product (MVP). Many analytics use cases evolve and move closer to the efficient frontier throughout their first few iterations. By starting off with the simplest version of your use case, you can be sure that you haven’t passed the frontier and overspent before your use case becomes operational.
Rule 3: Keep things as off-the-shelf and modular as possible. Don’t try to re-invent the wheel. In all likelihood someone has already developed a tool or a balanced mind+machine solution that can address the parts of your use case. Keeping use cases and their functionalities as modular as possible reduces complexity, speeds up development by way of re-use, and generally keeps solutions less risky and less costly over their lifespan.
Rule 4: Apply continuous improvement. Now that you have a working viable version of your use case, you should continuously strive to get closer to the efficient frontier. A successful approach is to have a team whose responsibility it is to prototype and implement improvement ideas for your use case portfolios. They can prioritize ideas together with their clients, trial improvements and hand use cases back to operation once they are in steady-state.
Moreover, these are the people who know what is going on across the company and can spot improvement opportunities, e.g., a specific automation for Portfolio A could also serve Portfolio B.
Rule 5: Document performance, learnings, audit trails and best practices. Knowledge management is key. Every part of every new use case should be documented in a way that makes it easy to find. Knowledge from previous use cases can inform your decision on whether to implement a use case or how much automation to introduce.
An important fact that many companies miss: proper knowledge management reduces costs because you can skip the development and testing stages.
The efficient frontier for any portfolio of use cases will become apparent over time. Just follow these rules and you will definitely see improvements in the performance of your use case portfolio.
If you are interested in reading more about how to employ the mind+machine approach in the most efficient way in your organization, click here.