A Whole Different League Data Analyst
I can safely say that no one around me know exactly what is it a data analyst do.
Some of you most probably are thinking of a person who works with a bar, a line, or a pie chart. Others describe a data analyst as someone who manages dashboards or reports. Well, that is how people usually perceive a data analyst: the number cruncher, the statistics machine, the dashboard wizard.
Although I also have no exact definition of what the job description is, through my experience as a data analyst, I discover a pattern among those whose considered successful. A hint, the formula is not only comprised of numbers.
Producing insight is easy, especially if you have a large amount of data (yay, big data!). Just take a significant amount of numbers, deliver the distribution with the trend and then — voilà — you have got something to present. However, whether the insight is actionable or not is a different matter. Even worse, there is the confirmation bias; people tend to believe an insight that resonance to their beliefs. Data analysts can have the impressions that they have done great jobs delivering plenty insights that most of the stakeholder interested in, while in fact, they did not create any impacts on the decision making.
There is a gap between numbers and business decision, a gap which a data analyst is hopefully can help to bridge. This is where we need not only a number cruncher but also a business-minded data analyst. They are rare breed of data analysts who are willing and capable of looking into problems from the perspective of business, and apply the correct statistical approach to answer those problems.
Though becoming a business-minded data analyst is not easy, it is sure worth the effort. The result may not immediately apparent, but in the long run, we can definitely notice the impact. These are the 3 characteristics that make business-minded data analysts stand out from the herd:
1. Understand the benefit of a proper analytical method
Most data analysts are aware that the mean is not a good representation of data, yet only a few acknowledge why. Maybe you have heard a data analyst suggested using median, but what are the trade-offs? Why is it better to present the distribution instead?
As a data analyst, it is mandatory to understand why we have to leverage a proper analytical method. We do not need to remember all the statistics jargons. It is better to know just enough with a full understanding of how they can be beneficial for the decision makers. Moreover, we have to be able to convince people that gathering enough sample or doing an A/B test before releasing a feature is worth the hassle. We need to listen to the business user’s concern as well and be open for negotiation. In the end, a data analyst’s role is to state all the assumptions and minimize the risks. The decision is left to the decision makers.
In front of your stakeholders, position yourself as a doctor, not a police. If we become such a burden, i.e. throwing all the rules with no room for compromise, decision makers will find a way to go around us. Let yourself become the police for too long and you will realize that people are trying to ignore your consultancy altogether. On the other hand, if we can successfully convince the stakeholders how data analysis can help them, they will crave for data before making any decision!
https://xkcd.com/1445/
2. Have a business-oriented goal
Goal determines what we need to do and why we do things. It is obvious that a team would not work if each of the members has a different goal, even though they seem similar. That is why the performance of a data analyst should be measured in regards to the business goal, period. Of course, we can have a data specific goal such as availability of monitoring dashboard or reports. The point is to clarify why the report helps in reaching the business goal and whether the report is the most impactful. What is the point of creating 20 diagrams which do not contribute in taking a decision?
When we set the goal to be business-oriented, we are expected to understand the business context. If we limit our knowledge only to measuring p-value or doing chi-square test, we cannot identify the business need correctly. We need to be able to have a deep contextual conversation with business people about the business they do or else, we will waste our time finding an answer to the wrong question. In contrast, if we are willing to learn the business dictionary, we can combine them with our technical knowledge to provide suggestions to pinpoint the right question.
Learning is not enough, we must care for the business! When we learn enough about the business dictionary, we can contribute to a conversation with the business people. Care, however, is what make one a whole different league data analyst. We are not only capable of joining in a discussion, we could also initiate the conversation. When people know that you care about the same goal, they will open up about all their concerns and ideas. From this point, spotting what question to answer and which task to prioritize is much easier, and our impact then will grow exponentially.
3. Be proactive and over deliver
This step is probably the most practical: deliver more than what is requested. You may have heard that “under-promise and over deliver” is a bad advice. However, over-delivering is indeed a useful practice to some extent. In particular, this approach helps to push the quality of questions that we could answer.
Although being proactive sounds straightforward, it would be nonsense without a comprehensive understanding of the business context. More numbers and percentages are not always better, instead what matters is the depth of the analysis and the relevance to the business goal. Never do an analysis without knowing the purpose and believing that it is a necessary one.
How to be sure? Be proactive! Get familiar with the data and challenge every analysis we do. For every analysis, at least ask ourselves these questions:
- what is the underlying problem the business user trying to solve?
- does this analysis tell the whole story?
- what is missing?
After we choose to care about the business and be proactive, we will naturally over-delivering. Instead of answering that our number of active users is 1000, explain why 1000 is 1000, not 900 or 1100. How does it perform relatively to previous time periods?
It is obvious that the business user wants to find out which feature drives the most active users, then why waiting for them to ask? To dig even deeper, challenge the term “active users” itself: do we have the same understanding as the business user? Is it correctly defined?
To wrap it up, yup, technical and research methodology knowledge are important for a data analyst, but there is one trait that most data analysts usually overlook: care about the business. With comprehensive understanding, we can have a fruitful conversation with the business user, hence creating bigger impacts through the ability to identify essential questions. Fulfill the expectation by being proactive and over deliver. Enter a whole different league, the league of business-minded data analysts.
Special thanks to my colleagues for the photos, review, and inspiration!
