Shipping Process Analysis
👈 Part II-IV| TOC | Part III-I 👉
In this post (Part II-V), I will do some analysis on shipping process and its effect on user satisfaction on our ecommerce.
To start, let’s look at the purpose of our analysis :
- How many late delivered order in our ecommerce? Are late order affecting the customer satisfaction?
- How long are the delay for delivery / shipping process in each state?
- How long are the difference between estimated delivery time and actual delivery time in each state?
Background
Big ecommerce company like Amazon, eBay, Alibaba, etc have changed the way we shop. US eCommerce sales alone increased by over $50 billion from 2015 to 2016 and continues to grow today. eCommerce now accounts for over 10% of all retail sales. But the increasing popularity of online shopping comes with pitfalls. Many of new ecommerce just cannot compete with existing competition. So how can we as a new and upcoming ecommerce compete with existing popular ecommerce.
One way to compete is by optimizing every touchpoint in our ecommerce business process to increase customer satisfaction when using our ecommerce. Every single thing your customer experiences related to your brand, from how fast your page loads on your store to the branded shipping label and packing tape on your package is an opportunity to make your brand their favorite. Think like your customer and imagine every contact and situation you could possibly have with them from the beginning of their transaction to the end. Then find ways to make each of them the best they can be. Customer are very demanding and spoiled because the big online retailers have raised the bar impossibly high for many smaller sellers.
Thus there are many challenge to increase the customer satisfaction, such as market evolution, increasing customer fulfilment by other competitor ecommerce, faster delivery expectation, etc. In this analysis, I want to emphasis faster delivery expectation challenge so we can see the effect of delivery process on our ecommerce customer satisfaction.
For this analysis, I will only use order data that had been delivered or cancelled, which means I won’t include data from order with status : unavailable, invoiced, approved, processing. Also, I will only use order data that only include single customer — seller relation (not multiple seller in 1 order)
Distribution of Late Order vs User Satisfaction
There are only 6.9% late order from late Sept 2016 until early Sept 2018. What is the effect of those late order?
We can see that the ratio of not late vs late order are slightly decreasing for each increasing customer review score. It means that late delivered order are more likely to get bad feedback than on-time order (or early delivered order).
Delay for Delivery / Shipping Process
We can see that 12.55 days are the average days needed to deliver the package from seller to customer. From the states perspective, we can see that more developed states (like Banten, Riau and Kalimantan) have faster delivery time than other states like Kepulauan Bangka Belitung.
Difference Between Actual vs Estimated Delivery Time
Our ecommerce has a system that estimate when the order will be delivered. How accurate were the system?
We can see that 11.8 (~12) days are the average difference between actual and estimated delivery date, which is really bad for up and coming ecommerce. From the states perspective, we can see that again more developed states (like Banten, Riau and Kalimantan) have shorter difference than other states. This might indicates that the more number of order, the more accurate the estimation system.
Shipping Cost vs Number of Order
It is very interesting to see that the more often the order occured for certain states, the lower the shipping cost (except from some states like Kalimantan and Riau). This might indicate that it will be hard to get new customer from low frequent order states because the shipping cost are so pricey.
Insight
As final takeaway, here are the things that we need to look out for :
- User most likely to give bad feedback for late order. AVOID IT!
- The average difference between actual & estimated duration is ~ 11 days. While that difference might be a way to avoid late order, the difference is too big and some user might don’t like it.
- The shipping cost will decrease with the increased number of order occurred in a state.
Recommendation
Based on insight that we get from the shipping analysis, we can recommend few things:
- To combat with the big difference between actual & estimated date delivery, we can try to make a shipping / delivery duration estimate system with higher accuracy for user (using Machine Learning!!)
- The proposed system above can also be used to measure the shipping process performance (KPI) on our ecommerce.
