The original version of this article was written in Spanish by Matías Zapiola and translated into English by ChatGPT.
How to think about your strategy without being a data expert?
Currently, it is increasingly easier to collect large amounts of data. We have the tools, the people, the skills; however, the challenge—and what will truly make the difference—is being able to ask the right questions.
Once you have them, it’s time to understand if you can answer them, how you will do it, and for whom the answer is. I like to think of this process in three steps through which the data will have to pass:
- Sources.
- Transformations.
- Presentation.
1) Sources
Every project involves trying to achieve a goal, it could be selling more, understanding who you are selling to, or reducing your costs. For any of these cases, you need data that can solve the problem (because, after all, whenever we talk about a data project, we are looking to solve some kind of problem).
So the first thing you have to do is understand if you have the data to answer your questions. Sometimes this can be difficult, but a good way to do it is to logically understand how your questions work. Let’s look at some examples:
- Suppose you ask yourself: “What can I do to increase my sales?” First, let’s think about what data can help you answer that. Keep in mind that sales vary based on price and volume. So, if you have data on price and volume, you can delve deeper and analyze whether it is advisable to increase price or volume.
- Another question: “What are my customers like?” Understanding them may depend on having certain data about them: their age, where they live, what gender they identify with, etc. In this example, you may not know where they reside, and this could be a hindrance. In that case, you would have to redefine your strategy and base it on data that you do have. Your strategy could take two paths in this case: change the definition of customer knowledge or start collecting the missing data.
- It is very common to wonder: “What can I do to reduce costs?” These costs can depend on volume, price, or the different strategies you use, which often can be easily quantified (like when you buy in bulk, for example, you get a discount) and sometimes cannot (like when a supplier offers arbitrary discounts without a definite rule). So, in the first case, we have the business rule (the quantity and its associated discount), and we can make a decision accordingly, while in the second case, we need to know that, although my costs depend on the discounts, we don’t have a way to use this data to make decisions.
This questioning exercise involves something very important: understanding if we have the necessary data to proceed or not.
2) Transformations
Once you know whether or not you have the data, we move on to the second stage: understanding how to transform them. Transformation can involve simple modifications (manipulation, cleaning, etc.) or it can be a longer process involving various operations. Another aspect to consider is whether the transformations can be done automatically (such as summing all rows in a column, for example) or manually (labeling each row in a column).
Let’s think about this through some examples, ranging from the simplest to the most complex. Suppose you want to understand:
❓ How much did you sell last month? -> For that, you simply sum the price by volume.
❓ What was your best-selling product? -> Of everything you sold, what sold the most (product breakdown from the previous information).
❓ Who is your top customer? -> The same, but by customer.
❓ What is the average price? -> Total amount sold divided by volume.
The previous examples account for simple transformations, but let’s consider what happens in more complex cases, such as labeling rows in a table. For example, if you analyze your bank statement and want to label monthly expenses into categories like “groceries,” “Utilities,” “Social outings,” or others. Or if you analyze audiences and want to group them at your discretion. In these cases, there is no formula that can be applied or a repeatable action; instead, you will have to manipulate the data on your own (labeling, summing unique values, subtracting, averaging, counting, or other manual operations).
This exercise of thinking about how the transformations will be is important because it will help you understand which tool you should choose to work with. In the case of simple and repeatable transformations, you could use a tool like PowerBI, LookerStudio, or Tableau. For more manual operations where you have to intervene with the data, you might need Microsoft Excel or Google Sheets.
3) Presentation
Once you have your sources and transformations, it’s time to act based on your data. Here, it’s important to understand who will act on the data and how they will support their decisions.
If you are the one who will act on the data, you can stick with your visualization tool and spreadsheet, as you will know them thoroughly. This will allow you to understand the data at a granular level and know how the models and tools work.
In the case that you have to present your data to people in managerial or more strategic positions, you should be even more concise. In these cases, a presentation or an executive summary with few charts would be best—something that can be easily sent and has the conclusions ready.
The level of detail your data has will depend on the type of person who needs to consume them. An operations manager will not need the same data as a CEO, for example.
To better understand what each level of the organization needs:
Operational Level:
- Granular data. For example, in our sales forecasting exercise, we know the prices and quantities of each specific item.
- Dashboards and spreadsheets for working with data.
- Contextual information such as the business problem and probable solutions.
Tactical Level:
- Intermediate-level data. For example, in our sales forecasting example, it would be knowing the sales values of each specific item.
- Scenarios and solutions with notes and justifications.
- Dashboard, spreadsheet to understand the scenarios.
Strategic Level:
- High-level data. In this case, we could state that it is only necessary to know the Sales result.
- Scenarios with risk and opportunity assessments.
- Executive summary or presentation.
In conclusion, whether you realize it or not, you are always working with data: every project is a data project. So, always think about your projects in steps and with these fundamental questions:
- Do I have the data?
- How will I transform them? What tools will I use?
- Who will act on the results? How should I present them?
