It was May 2015 when the Market Intelligence Data Solution process migrated to eClerx.
During the first business review of this process, Andrew (the client) said: "We are being one of the largest e-tailers and suppliers of more than a million products. Therefore, remaining competitive in the market is our primary aim. Also, eClerx should be able to meet the maximum product coverage without any quality issues in a cost-effective manner as we will also have other solution providers with whom we are in discussion."
Andrew's expectation was a challenge, as our team was struggling to deliver a little above 30% coverage with 70% accuracy. If we have to maximize, it meant revamping the entire process end-to-end.
Post returning from the business review meet, Saurabh (Senior Manager) called for an urgent meeting. He said, "Folks, we have a task at hand. The client is not ready to accept anything less than 85% product coverage and 95% accuracy."
"Himanshu…I want a plan of action immediately, we can’t lose this business ", our senior manager re-iterated.
My team along with a Black Belt got into a discussion and decided to take up a DMAIC approach to tackle this problem. In the first meeting itself, it became quite clear that this was going to be a challenging project to crack as the problem was deep-rooted and inherent to the process.
He said, "The problem at hand is the cumulative effect of multiple problems that are inherent to the process."
He conducted a brainstorming activity with our team to identify the various causes; it was like opening a Pandora's Box.
It was now understood that how deep rooted this problem was.
My team decided to attack this problem in phases. In the first phase, they came up with automation for each of the extraction, transformation, and loading processes. It improved the coverage and quality of the match to 70% and 90% respectively.
The client was informed to which he replied "You guys have done a superb job but a lot needs to be done. Keep up the good work"
There was a sigh of relief in our team. For the second phase, the machine learning approach was adopted, which improved the coverage and accuracy to 85% and 95% respectively.
Lessons Learned
- We realized that by continuously innovating and improvising on the existing systems and solutions, most complex problems too can get resolved
- One should look at a process from improvement perspective right from the beginning and not wait for a problem to arrive to take improvement efforts
- Excellence is a journey