Fundamentals

Ecommerce and Epidemiology

August 6, 2020

If you can sell products online, you can be a successful epidemiologist.  

Ecommerce and Epidemiology share a surprising foundation that contributes to success in each.  

As Morgan Housel writes,  “A truth in one field often shines a light on another. So let me tell you a story about what cats falling out of buildings teaches us about businesses surviving Covid-19.”

Instead of falling cats, I’ll tell you how selling groceries online sheds light on reducing HIV spread.

I used to work on the finance & analytics teams for Jet.com’s grocery delivery business (RIP).  Our only focus: creating a customer-centric business that people would love.  We tested new ideas often.  Different copy, focus groups, NPS scores – we used every tool in our toolkit to track how customers navigated our site, and if they bought bacon with their eggs.  We’d even test different pound ice packs and cardboard box configurations to keep packages dry and cool in 90 degree weather.   Everything revolved around creating the best end to end online grocery shopping experience.  The customer’s reaction and satisfaction lay at the center of every decision we made.

Effective epidemiology works exactly the same way as our online grocery business did.  To stop the spread of any disease (those without a cure, otherwise you’d just use a vaccine) in the 21st century, you have to provide a world-class experience to the population.  We’re the ultimate customer.  I’m not going to get into how awful the customer experience for COVID has been in the US (long waits for testing, slow turnaround time, poor customer service, the list goes on and on).  Rather, we’ll take a customer-centric view of the world as an epidemiologist focused on stopping the spread of HIV.  

Retention

What does every Ecommerce business care about?  

Retention!  

We had a whole team at Jet focused on retention rates. Retention rates answer the core question: how often did our customers repeat shop with us?  If they did, how long did it take them to come back?  Did they order more groceries when they shopped again?  If they didn’t shop again, what could we have done better?  We had to answer these questions if we wanted to create a world-class customer experience (not to mention a business!).  

Ecommerce businesses see average retention rates around 30%.  Any ecommerce business selling a repeat good needs to maximize their retention rate if they want any chance to survive.  

If you’re an epidemiologist with a 30% retention rate, you have a serious problem on your hands.  Working with a disease like HIV that needs consistent treatment to maintain viral suppression makes retention essential.  

Retention in HIV Care has been explored quite a bit; it’s a tougher problem to solve than in Ecommerce.  One of the main issues, described in the paper Measuring Retention in HIV Care: The Elusive Gold Standard, is that “Despite the recognized clinical and programmatic importance of retention in care, there is no recognized standard measure for retention in care.”  

In epidemiology, calculating retention and improving it remains crucial because of important negative externalities – those customers you don’t retain have the potential to infect others.  Epidemiologists help customers lose something rather than gain something.  Regardless, both fields work to retain every single customer.  They just come at it from different directions.  

Of the course, the next major focus after calculating retention rates in both fields turns out to be: cohort analysis.  

A cohort analysis defines a group of customers with shared characteristics, tracking how their behavior changes over time.  It’s useful in uncovering characteristics of both your best and worst customers.  These characteristics range from the date of first purchase, to age, to participation in a specific event.  

At Jet, our “perfect” customer spent an average of $100 on groceries with each order.  Why?  Well for one, that average order value (AOV) played well with our financial models; the higher dollar value contributed to a positive bottom line.  From a customer experience perspective, it provided evidence that we could simulate an actual trip to a grocery store.  Higher dollar value = more items = representative of a weekly grocery shop = customers using us for their weekly grocery shop = strong retention rates.  We did everything we could to in our cohort analysis to find these glorious shoppers.  

Multiple factors of course play into each cohort.  For example, we dove into the month they first started shopping, if they used promo codes, their age bracket, gender, specific basket composition (back to finding out if they bought bacon with their eggs!).  We sliced these cohorts every way possible.  All tied back to the key goal of increasing retention rates.

An epidemiologist would find all of this surprisingly familiar.  Just replace grocery shopping with their HIV treatment program.  

They’d ask:  which customers (patients) return to care?  Can we find in our cohort analysis any positive effects of a program decision?  Did we open a new facility that enabled working men to easily access?  If we see a cohort of young women with dropping repeat rates, can we tie it to that recently closed treatment center?  How did retention rates change for a cohort of patients transitioned to a new drug?  

Answering these questions and understanding these cohorts enables the epidemiologist to continue to improve their program’s experience.  Just as we continued to refine our website search and our packaging experience at Jet.

Wrapping Up

Success in Ecommerce and Epidemiology rely on the same underlying fundamentals.  It’s sometimes surprising when insights and methods from one field can apply the same way in a completely separate field.  

This idea shows the importance of pulling on common threads amongst different industries.  You never know when insights in one can unlock new understanding in another.  The best example of this comes from the book, Complexity, describing an economist creating a new economic theory:

“He still had no clear idea how to apply all this to the economy. But he could feel that the essential clues were there. He continued to pour through biology texts all that summer.”

Leave a Reply

Your email address will not be published. Required fields are marked *