Sunday Snapshots (20th October, 2019)

Churchill, eCommerce fraud, Small warehouses, Causal, and Ramen

Hey everyone,

Greetings from Evanston!

It’s been a week of setbacks. I’m excited about what is coming my way in the next few weeks and am trying to take things one by one.

A tough week always makes me come back to things that I enjoy – reading and writing. In this week’s issue of Snapshots, I want to talk about:

  • Winston Churchill and setbacks

  • Detecting eCommerce fraud using clustering

  • Why small warehouses are a big deal now

  • Causal: A web-based modeling tool

  • And more!

Book of the week

This week, I re-read parts of The Last Lion: Alone by William Manchester. It covers the period between WW1 and WW2 of Winston Churchill’s life when he was in the political wilderness. It’s a book that I like to come back to over and over again because it covers essential themes of the human experience – heightened precautions after a conflict, how to recover from setbacks, and how to thrive in chaos. Churchill is a flawed character. A Victorian man who struggled with the reins of power slipping from not just him, but from the British Empire. He led an extraordinary life and this period of darkness formed the backbone of his exploits in the Second World War.

Here are my main takeaways from the book:

  1. Writing as an instrument to remain relevant: Churchill was relegated to the back benches of Parliament during the interwar period. Yet he remained top of mind in the hearts of Englishmen so when they wanted a leader to free them from the looming scepter of Nazi rule, they knew exactly who to turn towards. How did he manage to do that? Writing and publishing. During the 1930s, he wrote for a variety of news publications across the globe. This was his platform. An intangible form of power that can be difficult to appreciate. While others on the back bench were at the whims of politics and popular passions, Churchill developed a direct relationship with the world through his writing. He warned his audience of the dangers of the Gestapo and how appeasement was not going to work with a maniac like Hitler. Not only did writing pay his bills, it’s probably what kept him politically relevant.

  2. Developing diverse information streams: One of the most fascinating parts of the book is how often Churchill knew more about the machinations of Nazi Germany than Downing Street. He became a magnet for information through government officials dissatisfied with how His Majesty’s government under Baldwin and Chamberlain was dealing with the threat of Hitler. He built this network of information streams through his years in public service across different parts of the Government and by using his writing to attract like-minded people which included journalists and tourists who had come back from Germany.

Alone is a long read and Manchester can sometimes go into details that feel unnecessary. I would only recommend reading it if you are interested in Churchill or the interwar years.

Long read of the week

Detecting organized eCommerce fraud using clustering

It shouldn’t come as a surprise to anyone that there is a lot of fraud online, especially when it comes to eCommerce. There are promo codes, cancellations, gift cards – many points of attack for someone to “hack” the system. They are estimated to cost retailers over $100B every year.

It’s difficult to catch because:

Fraud detection is a time constrained process that must happen after an order is placed and before it is processed for shipping. This typically gives only a few hours to detect fraud and only a small fraction of orders can be inspected by human experts.

The key insight in this paper by researchers at Aalto University is that “frauds to be clustered at a significantly higher rate than legitimate orders.” They are clustered (are similar) along the following attributes:

  1. Customer identity: IP address, email address, etc.

  2. Delivery information: Delivery type

  3. Shipping information: Shipping address

  4. Payment information: Card numbers, card type, etc.

At the end of the day, the algorithm the researchers propose is not ideal. It flags too many legitimate orders as fraud. This significantly decreases the lifetime value of legitimate customers and therefore makes the business case for this algorithm difficult to justify. But it’s a step in the right direction. Creating buckets of attributes can have more signal than the individual attributes themselves and this paper shows how that principle can be applied to fraud detection.

Business move of the week

E-Commerce Driving Bigger Demand for Smaller Warehouses

When you think of a warehouse, you probably think about a large industrial space spanning the length of multiple football fields somewhere out in the suburbs. But with customers demanding ever faster delivery of their items, companies want to place their fulfillment centers close to urban centers.

The problem?

Large, expansive spaces are typically not available close to cities.

So retailers and fulfillment services are moving towards a more distributed supply chain consisting of smaller warehouses within cities. Space in these urban centers is limited, so retailers are competing with developers with housing and other commercial projects – driving the price of industrial space high.

It is a trend to keep an eye on.

Gesture of the week

I love stories like this. Someone who found @timcammm’s wallet sent a note through a series of £0.01 payments.

Here’s a challenge for you: Go the extra mile this week to help someone out. Let me know what you do by replying to this email. I’d love to hear your story.

Product of the week

Causal: A browser-based modeling tool

This summer, I tried out a new product called Causal by Taimur Abdaal and Lukas Köbis. It’s a web-based tool to simulate outcomes based on data. The target market for the product are folks who are looking for more robust simulation tools in Excel but don’t have tons of heavy lifting to do in terms of data processing. I played around with the product for a potential new business opportunity I was researching during my internship at SpotHero and enjoyed it quite a bit. They’ve added some new features like sensitivity analysis and variable grouping in the past couple of months that make it even more compelling for power users. It’s a great complement to existing Excel models due to its ease of use and visualizations. I’m excited to see where the product goes!

Meal of the week

Last night, I went to Ramen Takeya in West Loop. We had the Fried Chicken Buns and the Spicy Chicken Paitan. The buns were better than the ramen. Definitely worth a shot if you’re in West Loop.


That wraps up this week’s Sunday Snapshots. If you want to discuss any of the ideas mentioned above or have any books/papers/links you think would be interesting to share on a future edition of Sunday Snapshots, please reach out to me by replying to this email or sending me a direct message on Twitter at @sidharthajha.

Until next Sunday,

Sid