First of all, thanks to everyone who left me a positive comment on my article last week – you encouraged me to continue writing more about business intelligence. This week I’ll talk about metrics that can help you make better business decisions. The tools I am going to discuss will help you place your products and/or services effectively, whether this is in a brick and mortar or online store.
Here is the data we will be using:
- Product 1 is Nike 1250 and product 2 is Adidas 6510.
- There are 20 transactions in your database that include both of these products.
- The total transactions you have recorded are 200.
- There are 40 transactions that include Nike 1250.
- There are 30 transactions that include Adidas 6510.
The Rule Of Support: This rule tells us how relevant the sales of two given products are. For example, let’s say that you have an online store that sells shoes and you want to know how often 2 given products are purchased.
The formula to figure this out is:
Number of transactions with the 2 products/Total transactions
In this case, 20/200 = 0.1. This means your customers by both Nike 1250 and Adidas 6510 10% of the time. If you figure out what 2 products have the greatest support, it would be beneficial to put them together.
The Rule Of Confidence: This rule tells us how confident we can be that if a customer will purchase one product, they will also purchase another product. The best indicator of future behavior is past behavior, so we need to figure out what the confidence level has been based on the past purchases.
The formula to figure this out is:
Number of transactions with the 2 products/Total transactions with 1 product
How likely is it that the Nike 1250 will be purchased when the Adidas 6510 is purchased?
= 20/30 = .6666 = 67%.
How likely is it that the Adidas 6510 will be purchased when the Nike 1250 is purchased?
= 20/40 = .5 = 50%.
-If you have a brick and mortar store, and you know that the when a customer purchases product A, they are very likely to purchase product B, it would help to ask “Would you like fries with that?”
-If you have an online store, you can setup it up so that when a customer is looking at Nike 1250, and you know that there is a very good chance that they will also want to purchase Product A with it, offer it to them at checkout.
The Rule Of Lift: lift tells us how much more or less likely a customer is to purchase one product when they purchase another product.
To calculate lift =
Confidence/Frequency of 1 product
We figured out that the confidence an Adidas 6510 will be purchased when the Nike 1250 is purchased is .50. The frequency that only Nike 1250 is purchased is 40/200 = 0.2
So 0.5/0.2 = 2.5. This tells us that a customer over twice as likely to buy Nike 1250 when he or she buys Adidas 6510.
You can use lift to figure out how much affect one product has on another. This information can be particularly useful when deciding which products to keep – if one product is giving high levels of lifts to multiple products, it might be a bad idea to get rid of it even if your profit margin is low. This is not always the case because obviously there are many more factors to be taken into account, but it can help make educated business decisions.
It is also important to note that lift might not be accurate if the database you are using is small. Confidence can be affected too much if the frequency of a certain product is extreme. Support is proportional to your entire dataset.
If you attempt to figure out the support, confidence and lift, and you notice something odd, it might be because of the notes above. Nonetheless, if the data-set is large enough and you have a good grasp of your products, you should try this out! These metrics are not limited to products; they can definitely be used in service industries.
Let me know if these metrics help!