Breaking Down Barriers: Using ChatGPT to Explain Data to Your Client
What is the hardest part of my data analyst job?
Not performing data analysis. Not learning new systems.
It is explaining technical information to stakeholders without a tech background — whether it’s students, coworkers, or board members.
When I assume a coworker has technical knowledge (e.g. that they know the difference between an average and a median), I see their eyes glaze over, and I know I’ve lost them.
Luckily, ChatGPT can help! Many of the prompts below can be used with the free ChatGPT membership. Certain options (like custom instructions and the Code Interpreter) require a ChatGPT Plus membership ($20 per month.)
Below are real-life examples of ways I use ChatGPT to explain data concepts to people without a technical background. If you aren’t interested in ChatGPT Plus, you can skip the first part!
First, set up custom instructions in ChatGPT Plus (accessible if you have a Chat GPT Plus membership.)
ChatGPT Plus has a new “Custom Instructions” feature, where you tell ChatGPT what it needs to know about you so it can provide you with the best responses for your needs. This means you’ll spend less time tailoring each ChatGPT response for your use case.
For example, I will not have to tell ChatGPT in every prompt that I want the response to be polite, short, and explain technical concepts in an easy-to-understand way. Once I have Custom Instructions set up, ChatGPT will carry these instructions through all my chats.
How to set up custom instructions:
- Go to the bottom left-hand side and click on your name and profile picture.
- Click on “Settings & Beta”
3. Click on “Beta features”
4. Toggle on “Custom Instructions: Try a new feature that lets you share anything you’d like ChatGPT to consider across its responses.”
5. Click on your profile picture again. Now there will be a speech bubble icon above Settings & Beta, called “Custom Instructions.” Click on it and set your custom instructions.
6. Enter your Custom Instructions and click “Save.”
7. Tell ChatGPT what it should know about you and how you would like it to respond — e.g. formal/casual, concise/long. I told ChatGPT to summarize data, analytics, and technical subjects in a way that a non-technical coworker would understand.
Next, let’s consider some real-life examples from my job where I explain statistics, tech, and data analytics to clients without a technical background.
Scenario 1: Your client is confused by your data analysis of employee salaries. They ask: “What is the difference between an average and median salary?”
ChatGPT Sample Prompt: “What is the difference between an average and median value of a dataset? Explain in terms that a person without a technical background would understand and use easy-to-understand examples.”
Note that if you have Custom Instructions set up, you won’t need the sentence about explaining in easy-to-understand examples, since this preference will be carried on throughout your chats.
I appreciate ChatGPT’s distinction between the average and the median, although I do have to say, the example at the end about splitting a pie is not very helpful.
When you’re writing an email to your coworker, you can take whatever you find helpful from ChatGPT’s response and leave the rest. If you are using examples with numbers, make sure ChatGPT calculated the average and median correctly.
Let’s say I didn’t like the pie example and want to try it again. You can change the prompt in ChatGPT: “Can you explain the difference between mean and median for an audience with a tech background?”
Scenario 2: Your client is still confused by your data analysis of employee salaries. They ask: “Why is there a difference between the average and median salaries? Is that to be expected? What does it mean?”
ChatGPT Sample Prompt: “What are some reasons why the average and median values in a dataset may be different? Explain in terms that a person without a technical background would understand and use easy-to-understand examples.”
Scenario 3: Your client wants to report on either the trimmed mean or the regular mean — they cannot decide and do not know the difference between them.
ChatGPT Sample Prompt: “Explain the difference between a trimmed mean and a regular mean. Explain in terms that a person without a technical background would understand.”
The first time I entered the prompt in ChatGPT, it incorrectly calculated the trimmed mean in its example, so I asked it to correct the trimmed mean. This is a helpful reminder to double-check any output from ChatGPT.
Otherwise, this is a very helpful and concise summary of the difference between the regular mean (average mean) and the trimmed mean. If you want to delve into the subject further, ask ChatGPT for examples of situations where the regular mean or the trimmed mean would be more useful.
Scenario 4: Your client decides to report on the 90th percentile, but when you provide them with the 90th percentile, they look confused. They ask, “What does this number mean?”
ChatGPT Sample Prompt: “What does it mean if a salary is in the 90th percentile? Explain in terms that a person without a technical background would understand.”
I love this ChatGPT explanation of the 90th percentile. It’s concise, straightforward, and gives a relevant example. If you want to delve into this further, describe your dataset to ChatGPT and ask it to explain what the 90th percentile means for your specific dataset.
—
I hope this article helps you as you explain complicated data analyses to coworkers, clients, bosses, or board members. It is crucial for data analysts to be able to explain technical subjects to people without tech backgrounds — because let’s face it, most people don’t have tech backgrounds!
Sources:
