In my previous post, I argued that brand is the predisposition of a group of people to buy.
You want to increase the number of people and their predisposition to buy to make your brand stronger. We should start by dividing people into two groups. People who have already bought from us are Current Customers. Everyone else are Potential Customers.
In this post, I am focusing on Current Customers. As you know, not all customers are equal. Some have bought from us dozens of times, some just once. Some buy new products at full price and some are discount hounds and the worst rip us off through fraud.
Regardless of type, each customer represents a forward cash flow (even negative). The better we understand our customer base and their cash flows, the better we can:
- Plan and buy inventory i.e. convert capital into sales
- Drive profits i.e. convert sales back into capital
- Acquire new, high value customers
Each of the three points above are points of leverage or compounding in the business. Nailing all three means more durable and faster growth and increases in our valuation and maximizes our chances of success. So the stakes are high and the rewards huge.
What we really want to know about our customers is:
- Which customers are truly high value?
- What traits do high value customers share?
- What actions can we take to create more high value customers?
- What actions can we take to generate more from our high value customers?
- How long do we keep high value customers buying from us?
- How good are we at acquiring new high value customers?
What tools do we have to answer these questions? I think here is where most people would say cohort analysis. When commerce people say ‘cohort analysis’ they usually mean calendar-based, first purchase cohorts where a customer is put into a cohort of other customers whose commonality is the month of their first purchase. For example, a person making their first purchase in June 2024 will forever be in the June 2024 cohort and all their subsequent purchases will be associated with the June 2024 cohort.
Cohorts are simply groupings of customers by various attributes. The grouping requires a commonality that all people in the group share. The logic in the analysis is that all these people share this commonality and then as a group, they did X, Y and Z so therefore we can predict how much of X, Y and Z the group will continue to do into the future.
Calendar-based first purchase cohorts help you:
- Understand if average purchase frequencies are increasing or decreasing over time and the rate of that change.
- Understand if sales from a specific cohort are increasing or decreasing over time and the rate of that change.
- Discern the effects of specific events on the subsequent purchase behavior of people e.g. a big sale or Covid or a stock out or introducing of a new product
- Predict the seasonality of purchasing behavior
- Understand average customer lifetimes and the decay in purchasing or the retention curve
- Predict purchases from customers you plan to acquire in the future
While a solid tool and analysis, calendar-based first purchase cohort analysis (for the sake of brevity, I will refer to them just as ‘cohorts’) has two important limitations.
First, calendar-based first purchase cohort analysis places primacy on the month of first purchase. It is saying that the most important thing about this group of people is that they made their first purchase in May of 2024. In SaaS and subscription businesses, this month of first purchase is the most important factor because customers have decay or retention curves and you must know when the clock starts ticking to anticipate these decay curves. Also many SaaS models have various renewal timeframes, so the business must absolutely know when contracts began so it can anticipate the likely cost and timing of the renewal or churn. If you have a subscription model, all this is true for you. But if your model is not subscriptions i.e you are an apparel brand or non-subscription food brand or skincare brand, is the month of first purchase the most important commonality?
Second, by definition, calendar-based first purchase cohorts contain an unknown cross-section of all types of customers. You have high value and low value, high frequency and one timers and the fraudsters all mixed together. Their single commonality being the month of their first purchase. Each cohort is thus giving you the average performance of that unique cross-section of the customer base.
Let’s go back to the questions we want to know about our customers and see what cohort analysis can tell us.
1/ Which customers are truly high value?
Cohort analysis isn’t much use here. We may spot differences in one month or another, but we don’t really know why.
2/ What traits do high value customers share?
Unless it’s the month they made their first purchase, we don’t have much to go on.
3/ What actions can we take to create more high value customers?
We can’t go back in time and acquire more of those awesome November 22’s. Did we do something in November 2022? Anyone remember?
4/ What actions can we take to generate more from our high value customers?
See 1/. Who are our high value customers again?
5/ How long do we keep high value customers buying from us?
OK. Cohorts are definitely helpful here. We may not know who our high value customers are. But we at least have a good sense of average lifetimes and decay curves.
6/ How good are we at acquiring new high value customers?
Who knows? Who cares? What’s the ROAS?
Calendar-based first purchase cohort analysis definitely has value. See the list of strengths above. And having a well understood methodology for predicting the behavior of customers we have yet to acquire is absolutely useful for long term financial forecasting.
But to really answer the questions that I believe matter most about our customer base, we need to add a different tool to the toolset - decile analysis which I am going to cover in my next post.
Previous post: Defining Brand: https://elephantherdconsulting.com/blog/blog-database/defining-brand