FECS Part 3: Customer Retention

Are people just here for the ‘gram?

Disclaimer: Customer retention is in itself a massive subject, and as much as I’d hate to admit it, it is sometimes an art figuring out what to track and how. As a result I’m going to discuss why it’s important.

Before we get to the halfway point, here’s a quick summary of what Part 1 and Part 2 told us about finding our best customers. They are:

• Unpack your averages to figure out what segments of customers are sending you not only positive vibes, but also cash money.
• Oftentimes, only ~20% of your customers in the top segment will be providing the most impact to your company.

So you’ve learned the basics of kung fu… now what?

Parts 1 and 2 provide great frameworks to figure out the who, where, and how that go into purchasing. As for the what, when, and why? For now, let’s dive into the when.

So let’s say I run ~Chiyoung’s Widget Factory~. I run through the subsegmentation exercises in Part 1 to find that cat-owning home chefs who live in Cambridge, MA purchasing through my Shopify store is the most valuable / engaged segment of my customer base, and I’ve identified the top 20% or so that purchase the most. Great. So what now?

In thinking about the various dimensions of our customers, we don’t know anything about their purchase timing. We need to investigate the when of our customers’ purchases.

Why is this when important? It’s important to use temporal purchasing data to have a continuous, evolving sense of what is happening, and compare to a previous baseline as the company develops.

Practically, purchase timing helps you build a model around customer behavior based on their purchase timing. It informs things such as customer classification, marketing timing, and loyalty programs.

How do you operationalize this? Is there a magic rule of thumb to make your business blow up overnight?

Yes… and no? Not really.

Long story short? No. But you can use insights from purchasing trends to point you in the right direction.

Consider the following three examples constructed using Extremely Illustrative Data that show product sales throughout the year (summing up to annual sales):

This is Extremely Illustrative Data. Note the spike around the holiday season compared to the relative low volumes in the rest of the year

This first one depicts a more gifting behavior-trending seasonality that peaks in holiday gifting season.

Again, this is Extremely Illustrative Data. Note the consistency of purchase volume — this adds up to relatively similar revenues compared to our spiky example.

This second one has practically no seasonality. There is a flat baseline of purchase all year.

A third set of Extremely Illustrative Data that shows a more hybrid combination of the spiky and consistent examples. Again, these sum up to relatively similar revenue totals.

This third example is more realistic, and has a combination of both seasonal spikes and consistent baseline purchasing.

The takeaway there is that there is a subset of customers who may be frequent purchasers throughout the year, while there are customers who purchase the product as gifts. This dichotomy already gives us two distinct customer groups to explore, and because we have identified this potential behavior we can start testing more targeted marketing campaigns (such as consistent refill subscriptions vs. Christmas gifts).

In the realm of having a continuous pulse on the business, comparing these customer groups across time can give insights on the effectiveness of your customer strategy.

For example, let’s use the last Extremely Illustrative Example from the previous section. Once you identify the consistently purchasing group and the gifting groups, you can start creating metrics within those cohorts such as average order value, retention of customers, and number of purchases a year.

If you looked at the average order value of this cohort and see that it grows year-over-year or month-over-month, you can be pretty sure that there’s something happening that is incentivizing your customers to keep purchasing. Similarly, if this value were shrinking, this provides good direction into diagnosing the root of this issue.

Having a specifically defined group and periodically measuring specific metrics opens you up to a more methodical, scientific approach to customer strategy. For instance, you can begin A/B testing between customer cohorts.

Let’s take a quick breather here. We’ve established the following:

• Temporal purchasing data allows you to have a continuous sense of what is happening in your business, and compare to previous baselines as the company develops.

And we’ve established why it is important:

• Operationalizing and even getting started by tracking a handful of numbers points you in the right direction.
• Having a continuous method of tracking metrics allows you to begin quantitatively measuring some kind of impact and effectiveness of strategy, as well as setting you up for more methodical, rigorous methods of driving customer strategy.

I can summarize everything in this section with one word: iterate.

Let’s play the tape forward. You’ve established your guidelines for defining your customers, set up your customer cohorts, figured out the metrics you’re tracking, and compared your metrics over time with baseline numbers and the numbers are killing it.

What do you do next?

We take our guidelines and baselines and throw them out the window.

Did I really just say that? Throw our hard work out the door?

Okay, maybe I was exaggerating. However, take a look back at what we’ve built. Fundamentally, our system of define customer guidelines → establish customer cohorts → define metrics → measure and compare metrics is now an engine we can reasonably rubber-stamp across different scenarios. Congratulations! We’ve stumbled upon rudimentary customer strategy analytics.

But why redefine guidelines or baseline metrics?

There are three reasons:

1. As your company develops, you’ve ideally grown your customer base, especially your top customers or superfans. This additional growth provides us the opportunity to slice and dice your superfan group even further, to find super superfans!
2. You’ve developed a new strategic capability. Beyond using this to dig deeper in your burgeoning customer groups, you can also use this to identify new customer groups such as “customers on the fence” or “emerging superfans.”
3. After getting to know your customers you may realize out of the handful of metrics you’ve identified, only a small number of them matter. It might be important to take time to jettison the metrics you don’t even look at anymore, contemplate new metrics to examine, or even split up old metrics (such as retention) into more granular flavors (such as net vs. gross retention!).

If you don’t continue to re-examine your cohorts especially as your company begins to grow, warning signs such as falling retention or average order value may get averaged out due to your burgeoning number of customers. As a result you may find yourself running into false positives and pitfalls, bringing you back to the issues we were solving for in Parts 1 and 2.

Iterate, iterate, iterate!

That was a hell of a ride. It might take a couple of read-throughs — if it helps to see how FECS parts 1, 2, and 3 mesh together, this piece may be worth reading.

This takes us to understanding the who, where, how, and when behind a superfan! The next part, Inventory Turnover + Hero Product, will dive into the “what” of a superfan.

And finally, a small announcement. I will be leaving Cue Ball to go back to school, so the wonderful Amy Tan will be picking up the reins and generally being the life of the party. Stay tuned!