Establishing Trust when Launching the Kindle
Data related findings, even sound ones, will not be implemented without trust among data engineers and business stakeholders. Data professionals must earn trust to be impactful to the operation. The following narrative illustrates the role of trust in finding meaningful solutions at Amazon.com
At one point, I worked in customer service operations at Amazon.com. At the time, Amazon’s annual revenue was ~$35B compared to today’s $177B+. Thus, Amazon was still learning to scale and launch products. Specifically, I was charged with leading a support team for the release of the Kindle Fire. Senior leaders felt servicing the devices and supporting the accompanying app store were of paramount importance. Establishing customer service processes including returns and refunds were part of my responsibility.
Our stakeholders were confident the addition of a tablet and an app store would result in improved brand recognition and revenue growth for Amazon. However, a mystery developed a few weeks into the tablet launch. My agents anecdotally shared stories of Kindle app refunds totaling $1000 or more. As their manager, I had to verify if the problem was erratic or representative of a larger product flaw.
My cursory investigation helped me understand that large refunds were almost exclusively based on in-app purchases. For context, one of Amazon’s key innovations was one-click ordering which was patented in 1999. Top business schools even cite this innovation as a “game changer” for e-commerce. One-click ordering truly was a game changer for the newly launched tablet without parental controls. We found that parents would hand the tablet to their children. The child would open a game application and begin spending “digital currency” within the game. Parents were unaware the digital currency implemented the one-click feature thereby charging the account real money!
My inquiry had anecdotally identified a problem, but Amazon is a notoriously data-driven company. We needed to present data along with the narrative to earn the trust of executives. Each week, using the app store data, my team compared the applications with the highest in-app purchases to their corresponding refund rates.
We would have been justified advocating for major changes to one-click or even stopping in-app purchases until engineers could solve the problem by adding parental controls. However, doing so would be met with great skepticism. We understood the executives were motivated by revenue growth and meeting tight deadlines. We earned the trust of the executives by collaborating on short- and long-term solutions not by suggesting extreme measures, even if those measures were justified in the data. The short term solution, which did not derail growth or delay timelines, was to scrutinize and possibly remove the top 10 applications identified in the weekly report. Long term, we mandated that in-app purchasing behavior be considered in the upcoming parental controls.
Ultimately, we earned executive trust by understanding their motivations and aligning our data-driven suggestions. Never expect that data supported recommendations are unquestionable. Your audience may have non-data incentives. Only through a mutual trust will lasting data-driven cultures thrive.
