How Product Analysts Can Use ChatGPT To Boost Efficiency

ChatGPT gets over 96M visitors per month and is more than 85% accurate — it can even write flawless code snippets. Despite that, not all people, including data analysts, use ChatGPT in their work tasks. Mostly, they lack understanding on how to do them both effectively and with minimal time investments.

Here are the tips with practical examples I got on using ChatGPT across various data analysis tasks:

Tip 1. Use ChatGPT for stronger communication

ChatGPT is a good ally when you need to streamline processes as it lets you strengthen the communication — whether you need to summarize experiments, draft technical requests for engineers, or compose emails. But you also need to remember that sometimes ChatGPT can be too formal in its wording — to change that, you can use prompts like “Make it a bit less formal” or “Let’s tone down the formality a bit”.

In the example below, I use ChatGPT to structure A/B test results. I briefly describe them and ask ChatGPT to structure the findings for effective communication with a product team.

As a result, I receive a neatly written, well-structured text I can share with my team.

Tip 2. Use ChatGPT for unit economics estimations

By providing ChatGPT with necessary inputs, it can generate approximate estimations you can use after in project planning and decision-making processes.

In the example below, I’m calculating the unit economics for marketing expenses. I describe the current marketing expenditures, expenses on the sales team, and customer support related to acquiring new users. Additionally, if you give ChatGPT the average revenue per paying user per month and the average lifetime, it can calculate how many paying users you need to get per month to cover the expenses.

As a result, ChatGPT not only says we need to acquire 284 paying users per month to cover expenses and generate profit, but also demonstrates the entire calculation logic.

Tip 3. Use ChatGPT for support in SQL tasks

ChatGPT may not be infallible in generating SQL code from scratch, but it is helpful in SQL-related tasks. It assists in streamlining SQL development tasks like formatting codes, incorporating window functions or replicating existing logic. However, it’s essential to validate the suggested code with sample data after you get ChatGPT’s answer to ensure accuracy and reliability.

In the example below I ask ChatGPT to format the provided code and add explanatory comments where necessary, so that I can share it after with my colleagues and let them understand it easily.

As a result, I get a properly formatted and easy to understand code, which I can share with my colleagues. As for the comments, they can later be refined to suit your preferences — you can make them less formal and omit unnecessary details.

Another example — equipped with pre-existing Common Table Expressions (CTEs) and knowing how to address my task with the use of window functions, I outline the instructions for what needs to be done.

As a result, I receive a ready-to-use additional segment of code that fills the gap in my solution. Subsequently, I’ll thoroughly test the solution on a sample of my data to ensure that the code performs as required.

Tip 4. Use ChatGPT for Python coding

ChatGPT’s proficiency in Python coding surpasses its capabilities in SQL. I believe it's because of the vast training data sourced from online repositories. Data analysts can collaborate with ChatGPT to obtain Python code aligned with their needs — they only need to give insights into table structures and data relationships. However, it’s also better to validate code outputs — use sample data to minimize the risk of errors.

In the example below, I provide ChatGPT with the details of my data and ask it to create a distribution table and the construction of a bar chart to visualize the distribution.

As a result, I receive ready-to-use code that addresses the necessary tasks. In this case, full sample data is provided, but in other cases we can also describe table names, their fields and contents.

Tip 5. Use ChatGPT for making presentations

Crafting compelling presentations can be a daunting task, but with ChatGPT it becomes more manageable. It offers assistance in ensuring coherence and professionalism — whether you need to refine wording for slides or structure entire presentations. By providing text inputs, users can receive well-structured content suitable for presentation slides and enhance overall presentation quality.

In the example below, I provide ChatGPT with an overview of the data I need to showcase in a presentation. Additionally, I ask how these data points could be utilized by a product team.

As a result, I receive a structured text for a slide, along with recommendations on how this data can be utilised effectively.

As product analysts use ChatGPT more in their work, its flexibility and reliability show how important it is for product analysis. By using its strengths and fixing any issues, analysts can become more efficient and gain better insights in their work.