What is the difference between a data scientist and a business intelligence analyst? What about a data analyst? The difference between them is confusing and there is a good reason for that: there’s no generally agreed upon definition for any one of those. This applies to several other types of analytics related jobs out there as well. If you are looking at a “data analyst” job posting the only way to truly know the primary responsibilities is to read the job description as it can vary from company to company, or even department to department.
The glass half full view of that issue is that recruiters typically pack job descriptions full of multiple data related keywords, so if you are looking broadly you are unlikely to miss a role. The glass half empty view of that is that it often requires additional work to sift through job descriptions, and that’s because these roles have multiple keywords which can result in lots of false positives.
The lack of consistent position titles does not preclude the existence of clear categories that analyst jobs fall into. I believe that all analyst jobs fall into one of these five categories. The following are those categories and their required capabilities and responsibilities:
Strategy – Strategy analysts use data to develop or improve the strategies and tactics that a business uses. They typically work on projects related to new markets, new products, process improvement, marketing efforts, and cost reduction efforts. These analysts are sometimes called internal consultants and their work can overlap with the type of work that management consultants perform.
Strategy analysts are the jack of all trades of analysts. They must be comfortable with both statistics and coding because they will be responsible for the collection and analysis of their data. They must also have strong business and communications skills. In short, their job is to use data to find something that can be improved, determine how to adjust it, and then convince others that their improvement ideas are correct and worth the effort.
Modeling – Most of the time the term data scientist is used it implies a role in the modeling category. Modeling can mean a couple of different things depending on the situation, but it always involves making a model designed to predict the behavior of people or systems. This can mean statisticians designing models to evaluate the effects of interest rate changes, analysts who evaluate the effects of strategies, or web engineers who write algorithms to suggest movies you might like based on past activity.
Modeling is the most mathematically demanding of the analysis job areas. Whether the practitioners are PhDs or self-taught, their statistical knowledge needs to be a cut above other analysts. Modelers must have coding skills, and depending on the specific type of modeling, elite level coding skills may be required.
Modeling can also be highly regulated in a legal sense. For example, banks will often make customer targeting/marketing decisions automatically via models, so those models need to be approved by government agencies to ensure they meet legal standards such as fair lending.
Reporting – Most companies, and departments inside companies, will have analysts in reporting functions. They are the individuals who create the official scorecard of how the business is doing. How well did sales perform this quarter? How often did services meet their agreed upon metrics? How often were products delivered correctly and on time? These are all questions that are tremendously important to the business, and to the teams within, so it is important that everyone is working from the same fact set.
Reporting is considered an analyst job because it requires some coding and some mathematical skills. Reporting analysts will pull the data and develop the reports, recurring and ad hoc, that discussions on the business are centered on. They may build interactive dashboards for sales professionals and managers to explore results, and they may devise new ways to measure what is important.
Execution – Analysts who have execution roles are similar to project managers. They are the people who actually implement strategic or tactical changes. Project management skills are the most relevant, but analytics and coding skills are often required to successfully handle an execution role. Execution is often tasked with discussing changes with stakeholders, and in order to do that successfully they should understand the statistical basis of the changes. They must understand how all of the systems and processes work in order to make sure that changes function as intended. They may also be tasked with updating the code which implements the change.
Administration – Some organizations call their database administrators analysts. Good database administration is important to an organization for many reasons, one of which is allowing other analysts to perform their jobs effectively. In order to develop strategies or solutions based on data, data needs to be available in a usable format. Businesses in the modern world create data at an ever increasing pace (TED – data transforms business ), and in order to gain value that data must be stored and indexed effectively. Good administrator analysts understand the business and the analytics being run on the data in order to have models and analyses run efficiently and accurately without monopolizing the company IT resources.
How they all work together
A hypothetical and oversimplified example
Imagine a bank has a credit card marketing strategy in place that targeted existing customers with credit scores above 700. Strategy analysts have been asked to evaluate the business case of lowering the threshold to a 680 credit score. They comb through databases and find that there are 200,000 current customers with credit scores between 680 and 700, and 33% of them have credit cards with the bank compared to 50% of customers above 700. If expanding the marketing strategy can generate credit card signups similar to those of the above 700 segment, that would mean an additional 34,000 customers [(50%-33%) x 200,000].
The strategy analytics team will meet with the execution analytics team and describe the opportunity to them. The execution team will then begin discussing the idea with the other stakeholders to ensure the change is executed successfully and there are not unintended consequences.
This strategy change will need to be approved by management and several operational and financial risk committees or executives. The strategy analysts, or the execution analysts, are responsible for bringing this change to their attention, highlighting the benefits, and outlining the plan to control or mitigate risks.
The bank is seeking to expand its credit card base and wants to move forward with this change. The execution team will talk to the modeling team to see what effect it will have on the overall portfolio risk. The modeling team will adjust the models to determine the change in risk and possibly recommend additional revisions. For our example here, we will assume the modeling team sees significantly increased risk if household income is below $80k/year, but no issues with incomes above that threshold. Strategy analysts adjust the change to target only customers with household income > $80k/year and reduce their estimate for the number of potential new credit cards to 20,000 because of the income cutoff.
The execution team checks the data sources used to identify which customers receive credit card marketing and notices that household income is not present. They will then meet with the administrator analytics team to develop a plan to add household income data to the source data so that the strategy change is possible. The execution team or the administrator team will also change and test the code that chooses customers to receive marketing materials.
Next the execution team will discuss the proposed change with the reporting team. Management will want to monitor the performance of the new accounts separately to ensure the change is performing as expected. They will also like to see how many new credit cards are being issued from this change. The reporting analysts will develop the reports to measure the signups and performance as well as checking how this change may affect other current reporting or goals.
The banking example gives readers an idea of how the categories can interact and why so many categories exist. Not all companies will have all five of the categories as defined roles. Some companies, particularly smaller ones, will have one person or team performing roles that span multiple categories. Knowing what type of role that you are looking for will help you in your search, and help you demonstrate how you can succeed in that type of role.