Using AI to convert Raw Data Into Actionable Insights
Taking raw data and converting it into something you can use is a key strategic advantage that almost anyone can tap into. Previously, your trusted Data Analyst would be needed to create you a good old dashboard.
Working in the consumer goods industry, data is key when making big decisions. This must be one of the most powerful tools that has been unlocked in our new era of AI. The best part: You do not have to be a data analyst or a spreadsheet wizard.
You can instantly generate charts and explanations that help you and your team understand the data that you are working with. You can now do complex data calculations by just using natural language.
I have been using the data from my payment terminals to identify who my top customers are, and what they are buying. Which directly influences my category buying decisions. I want to show you an example, but I am going to change the data set to protect my customer data.
I am going to show you how we can build insights from raw data, that you can use for actual decisions.
Get your Data into a CSV format.
Get your data source into an CSV format and get that imported into Google Sheets. Best to use google sheets so that Gemini can work directly with you data with no copy/paste needed.
Always open Gemini from within google sheets.
This is the part where we start asking questions. I am going to share some of my prompts. Always start off with giving Gemini a persona and some context. Think like a programmer. With something as powerful as a Language Model, you need to tell it exactly what you want to do. Giving context and scope is the most important step.
“You are a data analyst. This is data from my payment terminals. Provide me the total purchase revenue and total number of transactions. ”
From the prompt above, the model now knows that it needs to analyse data, and it has context to what the data is.
We can see from the response that we had 733 purchases and this generated $ 5 408.10
Now lets look into what my average daily sales are:
Prompt:
What are my average daily sales over the period?
“What are my average daily sales over the period?”
Gemini now gave me and average daily turn over of $ 383.86 over the last 14 days.
I had some refunds over the past 2 weeks . Although not much, it could be helpful to see what card brands had the most reliable transactions and generated the most revenue.
“Which card brand generated the most sales? Which card brand had the most failed transactions?”
This where you can see why the original prompt was so important. The model knows the goal is data analysis, and now gave us a visual representation on the question, using its own initiative. Mastercard has the most sales, followed closely by Visa.
Since this data is from an automated vending channel, we do not know our customers. Lets see if we can get an analysis on repeat purchases and see how often this occurred in this period.
I could have told the model where to find the data, but I specifically wanted to see if the model could find other patterns or other tags to identify repeat purchases.
Out of 309 unique customers, over 46% of them are repeat customers. The model advised on short term repeat purchases, as well as long term loyal customers.
This is an excellent insight to extract from this data. There is value in bringing loyalty programs into this business.
A great next step would be to pitch a loyalty program to the companies leadership. Lets see how we can create a short summary, ready for a quick meeting with the companies senior leadership.
Prompt:
I will present the data to senior leadership. Create an infographic to help illustrate todays findings.
“I will present the data to senior leadership. Create an infographic to help illustrate todays findings.”
Together with bunch of infographics the model has given me a full summary of todays insights which can double as the meetings minutes.
“I want to pitch a loyalty program to help reward loyal customers. Build me a short and concise pitch i can present in this meeting with senior leadership. ”
Added some key takeaways to help me with my pitch to senior leadership. Reading the data, it would almost certainly lead to increased revenue. Big slam dunk achieved from a csv file.
Spending 30 minutes using Gemini as my own Data Analyst I have made a couple of profound takeaways for my small business. Takeaways that in the past would be lost in the data.
Based on the transaction data that would have previously gone straight into the “to do list”, we have come up with the following:
Actionable insights gained from 20 minutes with Gemini posing as my own Analyst.
1. Strong Revenue and Transaction Volume
Total Performance: The dataset records 742 transactions, generating a total revenue of $10,748.20.
Transaction Value: The average transaction value is approximately $15.21, with daily revenue peaking on May 8th at $664.99.
Consistency: Revenue remained relatively steady throughout the period, with a notable increase during the second week.
2. Dominance of Digital and Contactless Payments
Preferred Brands: Mastercard and Visa are the most frequently used brands, accounting for 325 and 261 transactions respectively.
Payment Methods: Near Field Communication (NFC) is the primary payment method (433 transactions), followed by Contactless Reader (CLS) with 291 transactions.
Minority Brands: EBT also shows significant usage with 123 transactions, while AMEX and EFTPOS represent a small fraction of the total.
3. High Operational Efficiency and Key Locations
Success Rate: The system demonstrates high reliability with a 94.6% success rate (702 successful transactions). Only a small number of transactions (22) were flagged as potential failures (Status ID 250).
Top Locations: Stanhope is the highest-earning site, contributing $1,915.62, followed by Forrest Way Shopping Centre at $1,330.65.
Machine Performance: Machines at Lachlan's Square and Pemulwuy Marketplace also showed strong performance, each generating over $800 in revenue.
AI wont be taking away our jobs. Its changing the way we work. Look at AI as a tool to supercharge your thinking.
