On the face of it, the study of comparative politics and data might not have much in common. However, for Tess Merkulova, lead performance analyst at Captify, the similarities are so clear that she has used the former as an academic foundation for the latter.
She said: “At the heart of it, comparative politics tries to establish a relationship between a number of factors that lead to an outcome through either quantitative or qualitative methods,” she said. “It’s a study of A/B tests to a certain extent.”
"Once you have the basics, you can go in any direction."
The postgraduate degree gave her a theoretical base of available methodologies and taught her how to chunk a larger problem into smaller ones and find the best methodology to solve each one. In addition, she gained practical knowledge of R and other modelling techniques. “Once you have the basics, you can carry on in any direction,” she said.
Merkulova is not long out of university and so at times has to contend with fellow data professionals not taking her seriously because of her age. She deals with this by making sure that she is credible, not just in terms of the output she produces but in also terms of her network and general industry knowledge.
“Surround yourself with people you can learn from and with whom you can freely exchange ideas. Your work won’t always speak for itself. Sometimes you have to speak up about it or have someone else who will back you up,” she said as advice others who might find themselves in the same position.
On a day to day basis Merkulova collaborates with the internal engineering and trading teams, and looks for new ways to utilise data to enhance the performance of client marketing campaigns. She could also find herself working on roadmaps for testing and automation, meeting with demand side partners or building predictive models with her team.
“Everything I do is centred in data and what we can learn from it to do better. Data holds an incredible power if you know what you are asking it and how to get an answer,” she said.
Merkulova is a stickler for accuracy, as she wants to be sure that anyone using her model is able to make the best decision possible. Data quality and data validation are two issues she comes up against every time does any analysis or modelling. She said: “Recommending someone do the wrong thing is something I’ve always been terrified of doing. Now I always spend a large portion of my time making sure that the data or assumptions are correct. If you query the same database in different ways and get a different result, something is wrong.”
As such, she is a big advocate of understanding mistakes and fixing them permanently as it will benefit everyone in the long run.
Merkulova is also a fan of answering well-defined questions, as she said that without one it is very easy to get sucked into a black hole of data exploration or produce a result that doesn’t answer the core question.
There have been times when she has produced output that the was not what the stakeholders were looking for because the question was too broad. She said: “It is always worth sitting down and having a chat with the person who is asking you for an output to truly understand what both of you are trying to figure out.”