Automated business decision-making provides a significant competitive advantage for companies today because, generally speaking, the faster you can decide on the right response to an opportunity or risk, and consistently act on it, the better the business outcome will probably be. By automating key operational decisions, those day-to-day, repeatable decisions that run the business — like what loan application to approve or reject, what product to offer a new customer, when to trigger an agent to call back a client to assist, when to re-route cargo, etc. – companies can streamline processes, manage risk more effectively and increase profitability.
So, when should you automate an operational decision?
The easy answer: you should automate those frequently occurring, highly variable decisions that often need to be made in real time, such as:
- Location-specific decisions, such as eligibility verification or pricing
- Customer-specific decisions, such as sales authorization, priority assignment or contract-related provisions
- Product-specific decisions, such as configuration and availability
- Process-specific decisions, such as workflow routing, approvals and straight-through processing
But in reality there’s a continuum of automated decision making — from completely automated decision systems on the one end to purely decision support systems on the other. The former captures the expertise of business people and streamlines the frequently occurring, predictable and low risk decisions by completely automating them, while decision support systems help humans to make high stakes and unpredictable decisions by either simplifying or supplementing existing information.
The following table (Parasuraman, Sheridan, & Wickens, 2000) identifies several levels of automation in a system. When determining the level of automation your system needs, consider using this scale to score the level of automation and then look again closely to see if more can be achieved. Choosing the right level of automation is essential to project success and reaping the tangible benefits of decision automation.
| Automation Level | Automation Description |
| 1 | The computer offers no assistance: human must take all decision and actions. |
| 2 | The computer offers a complete set of decision/action alternatives, or |
| 3 | narrows the selection down to a few, or |
| 4 | suggests one alternative, and |
| 5 | executes that suggestion if the human approves, or |
| 6 | allows the human a restricted time to veto before automatic execution, or |
| 7 | executes automatically, then necessarily informs humans, and |
| 8 | informs the human only if asked, or |
| 9 | Informs the human only if it, the computer, decides to. |
| 10 | The computer decides everything and acts autonomously, ignoring the human. |
(Parasuraman, Sheridan, & Wickens, 2000)
For example, an interactive sales guidance system that helps a call center agent find the best offers for a client when engaging one on the phone, is characteristic of an automation level 3, where the computer offers several pertinent product suggestions. Control is given back to the human to make the call on which of the shortlisted offers is the best in the given situation. If it were possible to truly come up with a single best recommendation every time, the sales guidance system could move up to level 4. The advantages here are that the client doesn’t need to be engaged further on the phone and the agent can sell to more customers, assuming of course that the optimum product selection was made by the decision support system.
In the case of an automated insurance claims processing system, the system is typically in the 8 to 9 automation level range. Most of the claims can be processed completely automatically and in some cases, the claims processing system may raise a flag on certain cases for human follow-up. With claims processing, the conditions are much better known and predictable, as the insurance company sets up the various possibilities, when compared to a much less controlled environment where a client must make a buying choice.
At the end of the day, a hard line doesn’t separate decision automation from decision support systems, but simply moving down the spectrum towards full automation can yield great benefits. BeneCard (IBM Corporation, 2010), for example was able to increase its straight through processing to completely automate over 80% of its claims and WorkSafe Victoria (IBM Corporation, 2011), a government agency in Australia is up in the 84 to 85% range for claims automation as well. These real life success stories show that while it takes some creativity and fortitude to be able to simplify complicated or hidden decision logic and then automate it, the payoffs for those businesses that do can be huge.
Works Cited
Parasuraman, R., Sheridan, T., & Wickens, C. (2000, May). A model for types and levels of human interaction with automation. IEEE Transactions on Systems Management and Cybernetics , 286-297.
IBM Corporation. (2010, September 24). ibm.com. Retrieved October 2011, 2011, from Benecard builds a smarter claims process with WebSphere ILOG JRules
IBM Corporation. (2011, May 5). WorkSafe Victoria — IBM Client Reference video. Retrieved October 12, 2011, from You Tube
