Recently, WNS Global Services sponsored a roundtable discussion in Atlanta, where pre-eminent new-age technology practitioners had gathered to share their thoughts on successes, challenges, and next frontiers. This group, that cut across industry segments, spent a fair bit of time discussing and debating not the alphas and betas, but how several good algorithms and techniques did not see the light of day – not because of their technical deficiency, but because the business did not sufficiently adopt them.
This caused me to reflect.
During the course of my career, I have had the opportunity of overseeing several implementations of decision-support systems, both, from developer and business side of the table. I learned the nuances (or should I say politics?) of successfully managing an implementation quite early in my career.
I had just joined a U.S. major airline and was tasked with developing solutions for flight traffic forecasting and fleet optimization using advanced operations research and statistics – yes, these terms existed before all got subsumed into Artificial Intelligence (AI)! My results had to be approved by the senior-most network planner in the department, Bill Thompson. My models needed to yield sufficiently different results and demonstrate improvement to management – but could not be so different from Bill Thompson’s manual approach that he would disregard them right off the bat!
We subsequently memorialized this gate that we had to pass through before eventual implementation, as the Bill Thompson Test.
My experience suggests there are five key ingredients to the secret sauce of adoption.
-
Learn the Ecosystem: Seeds for successful implementation are sown in nascent stages of the project life cycle. It is crucial to recognize the “Bill Thompson” of the business ecosystem, and the spheres of influence of key leaders. Almost always, the real drivers of success (or failures) are not the executive sponsors or their directs, but someone else in the organization. Find out who they are and work in tandem.
-
Learn the Business Speak: AI practitioners often get carried away by their terms, terminologies, methodologies, and forget that there is a business that could care less for these “exciting” techniques. Businesses demand output. It is imperative that the practitioner invests in understanding the intricacies of the domain and incorporates them adequately into the algorithms to ensure productivity.
-
Talk the Walk: At every step of the implementation life cycle, the practitioner should take the business counterpart (as defined in the first point) along, setting the right expectation, outlining the risks, and ensuring a soft landing. It is not sufficient to be adept at predictive methodologies, but also one needs to be able to predict the curveballs that could derail the implementation. Proper governance and constant informal communication go a long way.
-
Walk the Talk: With the right expectation set, it is now time to deliver! With your “Bill Thompson” sufficiently on board, don’t hesitate to take a back seat and let the business close the deal for themselves.
-
Post-operative Care: The days right after implementation will chart the course for years to come. Practitioners should stay joined at the hip with the client during this critical phase, quickly resolving issues and providing comfort that the product is of industrial strength and will indeed move the business forward.
Hope the next sauce is just spicy enough for your Bill Thompson’s liking!