The Hidden Value of Data-Driven Development

They had invested in an app to help employees submit their expenses simply by scanning or taking a picture of their receipts. The app was well-received and increased the rates of on-time expense submission, except… when they interviewed employees with low rates of usage, they found they were all iPhone users. 

As it turned out, the app was booting slowly for those employees, so while they were waiting for the app to boot, they often got distracted by other tasks.

Since submitting receipts wasn’t a high priority for these people, they just procrastinated submitting the receipt another day or week. Even more annoying for these employees was when they were expensing something on the spot—for example, the invoice at a restaurant. It was embarrassing and frustrating to pull out the phone and not have it immediately click and upload the receipt, especially if they were with other employees or a client. Then, when they came back to it later, they ended up wasting time later searching for where they had put the receipt.

Jesse, the Clevertech team leader working on the app was working according to a project plan, and the app booting time wasn’t first on her list of priorities. However, at Clevertech, all of our employees are trained to listen carefully to the client’s concerns. Even though technologists and business people don’t always speak the same language, when everyone cares about the success of the project, we can reach an understanding. In Jesse’s weekly meeting with the client, they brought up this as the primary issue they were dealing with.

As it turned out, this client felt they weren’t getting the results they wanted, and it was starting to erode the trust. Jesse quickly turned around the development team to focus on the immediate task of speeding up the start time, taking care to document their research and development so they could share the results with the client.

In this particular case, the answer had to come through rapid and accurate analysis of the problem. The customer didn’t have the tools to show Jesse what was happening, but at Clevertech, we use a combination of tools that allow us to get to the source, pinpoint problems, and fix them rapidly.

Jesse started with the APM and Kibana data, shown below. It turned out, of course, that the customer was correct and the startup time was slow. The first finding was that the average duration for calling the “SubscribeApple” function was taking 1,126 milliseconds, and for some users, the time was more than triple that: 3,592 milliseconds. That’s certainly enough time for a user to get distracted or frustrated and close the app. If they were just using it to photograph one receipt, well, it’s easy to see why they didn’t bother to wait for the app to boot up. In fact, as shown in the “Impact” column below, this issue has a very high impact on the overall operation of the application.

At this point, Jesse was able to show the before-and-after results to the client. In other words, the communication wasn’t just a feeling or perception on the part of the client. It was also based on speaking a different language. Jesse was looking at statistics that showed her success - but the client didn’t have access to those same tools. When she showed them the graph below, they could see in a visual way how the startup time had moved from that danger zone down to a consistently low time frame. This gave the client confidence to make an announcement in the organization that this problem was resolved, and ask iOS users to try again with the new upgrade

Data-driven development makes a huge difference in software projects like these, because people get wrapped up on their desire for immediate results. Optimizing a software app is done in stages and small shifts. This entire example was about the shift of one second in startup time, which is not necessarily perceivable to a human - particularly if they are biased to look for a slower start time. 

Similarly, trust is built up over time, building proofs that a team can deliver, that they tell the truth, and that they treat others well. A common language - particularly a data-driven language - builds that trust similarly. By showing the incremental changes on a week-by-week basis, service providers build the trust of those they serve. By viewing the same statistics, we can ensure that we are communicating about the same problems, or, if we aren’t, we can look at other ways to address the problem.

Creativity and data aren’t at odds with one another. Clevertech employs a highly creative team, but we work within the constraints of the real world. We train all of our team members in listening and communication skills, so that we can effectively address the most important concerns of our customers. Knowing the facts and seeing the data allows us to have the basis for our creativity. Measuring, monitoring and logging data over time provides the performance tracking that allows us to apply creative solutions to real world problems.

Working with Clevertech means you get a team that cares deeply about your project. We use the latest technologies and data-driven development to get results.

Reach out to find out more about why the top global corporations prefer Clevertech as their outsourced development team.

If you’d like to find out more about our culture at Clevertech and current job openings, check out our YouTube channel and visit clevertech.careers.

Disclaimer: The story is a fictionalized account to protect our client, but the statistics from APM and Kibana are real data from a real project.