David ..you are always the best . I did my Maters in Data Science in 2017 with your Titanic dataset Data Analysis with R . I love how you simplify concepts
Thank David for the video. There is an error in recency. It is ranked from most recent as 0 to 9 as the worst score. So the filter should be 0 for recency , 9 for frequency and monetary. You should get 82 best records. cheers
Hey Dave, super simple tutorial. Helpful. I see that you have divided it into docile. I want to use the quintile format. Could you suggest the formula for the same?
Hey Dave. This is a good take on the RFM model and its application. I had a question though, shoildnt we reverse the current tiles for recency? I believe if one is a more recent customer, then one must be rated higher
Yes, this is a bug in the original video and Excel file - apologies! 🙄 I made notes of this in the video description and comments. The Excel workbook in the GitHub has been fixed.
Hi David, do companies use Excel or SQL or any of the other languages for performing RFM analysis? Moroever, how often are we supposed to perform RFM analysis (for instance the Telecom industry)?
Doh! That's a bug. 🙄 The recency score should be higher for more recent purchases. I shall fix the Excel workbook in the GitHub and make notes on the video.
Hi David wanted to know if you can help having into range of 1 to 5 (with interpretation into segmentation of Champions , Potential Loyalists, New Customers ,At Risk, Can’t Lose Them ).Will be helpful
I'm a bit weak on pivot tables (must look for a tutorial). Meantime could I get a similar result by sorting excel columns by a...then b... then c... ? Thanks!
Hi, I have a question about the customer selection for Recency. At 12:55, you chose segment 9, which identified customers with DaysSinceLastOrder greater than 300 days. However, if we select segment 0, it would capture customers with a lower DaysSinceLastOrder threshold, potentially including those who ordered more recently. So we might want to consider segment 0 instead of segment 9?
The regency score needs to be inverted if you are using 9 as a good score and 0 as unfavorable (quickly done with a vlookup up table). In other words, fewer days returns a low score using the percentage.exc calculation….so the 0 should really be a 9 to be consistent with high frequency and monetary numbers as being favorable.
Can you create a separate video on how to use RFM analysis in a business setting? for example, what are different customer sergmentations based on the RFM scores, and how would the business act on them? E.g what would you do after you identify power users based on scores X, Y Z? what would you do with people almost at churn (e.g used to be power users, but haven't used them in a long while)
⚠ NOTE - There was a bug in the original Customer Recency calculation. This has been fixed in the Excel workbook. ⚠ 💥 Learning R programming is easy for Excel users! 💥 📺 ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-MNpsyjSuR20.html 👩🔬👨🔬 Learn to use your basic Excel skills to analyze the business like a Facebook data scientist: 📺 ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-xIXymabyFIM.html
Hello Dave. I like your tutorial, it helps me so much doing my homework. But there's still homework that I can't do. So I have a question, can the results of the RFM rank/score be followed by making a cluster of k means? Thank you in advance.
If I understand your question correctly, you can think of RFM as a clustering technique. For example, after performing RFM you then use exploratory data analysis (EDA) on the customers with 999 scores. You then perform EDA on the 111 customers to see if there are any interesting/insightful differences between the two groups (e.g., age, gender, education, etc.). This is a classic customer segmentation exercise. You could also perform a k-means clustering on your customer data to arrive at some clusters (e.g., k = 5 clusters). You would then perform EDA on each of the 5 clusters, again looking for interesting/insightful differences between the groups. This is also a classic customer segmentation exercise.
great, but your recency score should be inverted - 9 should be for the most recent order (least number of days since last order) so I would add: 10 - (your formula)
While it is certainly possible to implement the CART algorithm in Excel without VBA code, it would be very error-prone and complex. I would suggest learning to use R (which is very easy for Excel users) instead. Check out this video on my channel that demonstrates how Excel skills makes learning R easy: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-MNpsyjSuR20.html
