Error, Office Not Found - A Data Science Memoir

By: Beata Borzasi

My first memory of working at boberdoo dates back to 2014. As a Hungarian data science fresh grad I received a one-year grant to gain hands-on and relevant experience in the United States. boberdoo.com, on the other hand, was looking for fresh ideas for research and development in the area of data analysis. So here I was in Chicago, in a completely new world, far from home, in the lobby of our office building, nervous and curious at the same time. With shaking hands, I checked the list of companies on the wall to figure out where exactly I’m supposed to go. As I scrolled my finger down the long list of companies, all that stress turned into hearty laughter: boberdoo.com #404. “Office Not Found” - though it would have been a great joke to start with, I kept my cool and walked in with a big smile instead.

Data Science 404

The same day I was introduced to basic concepts, data points and all the important information of the lead industry. Being the first data person at boberdoo.com felt like someone bought me a whole playground, but after all my life I was only allowed to read about it. I got to decide whether I want to jump on the swing, use the spring riders or just spend the rest of the day in the sandbox. I saw, and still see endless opportunities for both boberdoo and our clients: customer profiling and lead segmentation, email optimization with text mining, bid experiments, profit margin adjustments using artificial intelligence, vendor network analysis, machine learning-based decision support systems - and the list goes on. 

In the past 5 years, most of the above mentioned smart solutions have been implemented. However, data being the oil of the 21st century, the list continuously expands, ideas grow in number and complexity, while requests from clients sitting on a data gold mine come in more and more frequently. The rules became simple. You either adapt and implement intelligent solutions or slowly fall behind and risk the future of your business. Today, what we once called innovation is more of a day-to-day necessity. 

While keeping the same enthusiasm towards working on custom developments tackling specific challenges of our clients, in 2020 we plan to put greater emphasis on packaged solutions and strive for generalized data science features made available for everyone that are not only cost-effective but have low entry barriers (eg. short checklist for system and business requirements), are easy to implement (eg. no need for major changes in the lead system) and can be switched on and off whenever needed. Through these solutions we aim to answer the most frequently asked questions of our clients: 

  • Do the changes I made in the filter setup have a major impact on my budget? Is my daily spending normal? Did I *%#&^%* something up? 
  • How much should I bid on certain leads to keep my CPA low? Which leads turn into a signed contract with the highest chance?
  • What profit margin should I use for certain sources? How can I make sure to fulfill my partners’ demand while keeping my profits high?
  • Which bid experiments should I run next? What segments show the biggest potential? 
  • Which leads should I target through our social media campaign?

AWS Marketplace offers a digital catalog with thousands of software listings from independent software vendors that makes it easy for our clients to test and buy solutions that run on AWS. Thus, becoming a certified AWS partner is definitely one of the most emerging items on our todo list. 

Despite the current and upcoming challenges in the industry, we’re looking into 2020 well prepared, with increased bandwidth and new ammunition. Zsolt Nagy, joining our data science team just recently, will strengthen our expertise in research and implementation. As a former Ph.D. student in Sociology he will focus on lead demographics, segmentation by different lead performance indicators and various predictive techniques. His experiences at IBM will help us better structure our projects and follow coding principles like other multinational tech companies.

Data science projects are considered successful only if they create tangible value for our clients hence for us to develop solutions that ease your pain-points faster, we need to understand your business thoroughly. With Zsolt on board, we’ll have the opportunity to allocate more time to consulting, data coaching, and understanding your needs through 1:1 sessions. 

With this we hope to not only meet current expectations but continue leading our clients on a path of growth, innovation, development and competitive advantage. Cheers to that!

If you liked this post, stay tuned for more! In the coming months we will have more Data Science Deep Dive posts including: system health dashboard, new features for bid experiments, AI based automations, bid suggestions and profit margins.

Learn More