“They reproduce – that’s how you know!”

During one of my business trips, I had an interesting encounter with a person responsible for maintenance of the fish tanks in the office. As we chatted about his work, he shared an insightful moment. His employer wanted to measure the results of his work and had once asked him, “How do we know you’re doing a good job?” He pointed to the fish tank and proudly said: “See those little fish? Fish in this tank reproduce in captivity and that’s how you know I’m doing a great job.”

In this post I would like to go over my experiences with measuring success and work results in the context of different jobs I had. I want to highlight some personal learnings from each of these experiences.

Experiences

In my first role I didn’t care about the business metrics or showing the results of my work, I was a grinding machine – I had a job and I had it done. Did my management care about the work I did? I bet they did, otherwise they wouldn’t put me through two promos. With my first experiences of leading a team I started looking at things like tasks completed, sprint burndowns, etc. In retrospect I was very focused on tracking “work completion” for myself and my team that I might have overlooked connecting this work to actual business impact. I was lucky someone did this for me and our work delivered results but if I was there now I would have spent more time understanding the business.

At my next job, my team owned multiple backend services where performance metrics were critical. “We need to handle 7TPS on a single node, how many nodes do we need to handle the entire traffic?” “This API is slow and has weird behavior of being very slow for some of the requests. Can we figure out the bottleneck and get P99 of requests <150ms?” I think this was the job when I truly understood how technical solutions connect to business success. I found the performance bottleneck and this allowed for a “bet from autocomplete” feature launch which improved user engagement by xx%. (Btw, the bottleneck was a forceful garbage collection as a way of reducing memory, instead I got rid of this memory optimization and instead asked for more memory on servers hosting the service).

Things were somewhat strange at my next job. At first I worked on an app for IAEA agents to use to record their activities during their visits to nuclear facilities around the world. I didn’t know exactly how they measured their success even though I was curious. But later I had a chance to work on a scientific/statistical tool that would take raw nuclear material accounting data and perform all kinds of analysis of the data, come up with reports and KPIs. In a way this was a measurement product by itself.

Amazon is famous for being data driven and measuring everything. When I worked there everything had to be backed by data. And this wasn’t just for engineers. Same standard applied throughout the entire company. My promo had to be backed by the list of all the projects I completed, impact they had, stats on code, etc, etc. Dates mattered a lot as well. Market is ruthless and requires things to be completed before the deadline (think Prime Day, Re:Invent).

Google is another example of an extremely data driven company and I have had a chance to work on things related to how Google makes decisions based on data. Effect of any smallest change is measured on a small percentage of traffic in order to make an informed decision. There is an experiment almost behind any code change. Google taught me to always add a section “Landed Impact” at the end of every design doc and then populate it post launching the document with specific realized impact. Google had so much impact on my life that I’m setting my personal goals using OKR methodology. If you want some related reading, consider reading the book “Measure what matters”.

Conclusion

Understanding the success of a project is about aligning technical results with business outcomes. For example, if you’re optimizing code for faster processing, link this improvement to reduced server costs or increase in revenue, and then measure all of these components. And from a career perspective, you always want to be able to say what you did and what impact it had. The impact has to be stated in numbers and should have some meaning behind it. People should care about your numbers. If no one cares about your numbers, think again if what you are doing is worth doing.