A customer segment is a relatively homogenous group of customers that share important behavioral and/or demographic characteristics. One of the main concepts in the world of customer analytics is customer segmentation. Segmentation occurs in a variety of forms and levels of sophistication – from simple demographic segmentations to more sophisticated segments built according to business rules or based on scientific models such as cluster analysis. No matter how one arrives at these segmentations, one must understand both their composition and the differences between them. This post explores a methodology for segmentation discovery in three steps.
In order to explain these three steps in a more descriptive fashion, let’s use the analogy of a vineyard. Let’s imagine that I have a friend, a winemaker named Pierre Francesco. He owns a vineyard located on Mt. Edna in Sicily, and after each year’s harvest he conducts a broad analysis of the different grape varieties in order to manage his vineyard more efficiently.
Step 1: Sizing Up the Situation
Our friend Pierre Francesco grows three varieties of grapes: Chardonnay, Riesling, and Cabernet Sauvignon. After each harvest, he needs to measure the amount of grapes from each variety.
As we see in the following chart, Pierre discovers that Chardonnay grapes are the largest harvest, followed by Cabernet Sauvignon, and then Riesling. He also determines that there was an increase in the number of Chardonnay grapes harvested over the last few years, whereas the number of Riesling grapes over the past few years has remained about the same.
Segmentation by Grape Type
This is the first step in understanding segmentation: measuring the size of each segment. Over time, one may spot trends in the size of each segment.
Now let’s leave the vineyard for a moment, and take an example from the business world. We’ll examine segmentation according to different geographic regions, aimed to spot any growth trends over time:
Examining our segmentation from this perspective enables us to discover that customers from the United States and Canada constitute the highest portion in the database of the seven geographic regions. In addition, their relative portion of the database has risen over time.
Step 2: Discovering Segment Attributes
Returning to Pierre and our vineyard analogy: once we’ve sized up the situation with our grape varieties, we’ll want to dig deeper and ask ourselves what ingredients, or attributes, make up each grape variety. Pierre will want to measure the current level and trends of precipitation, altitude, soil, fruitiness, aroma, color, sugar level, etc. of different grape varieties to gain insight into the internal components that make each type of grape – and the resulting wine – unique.
As we see in the following example, Pierre measures the sugar level of each type of grape, in addition to measuring these sugar levels over time. He now understands that Riesling currently has the highest sugar level, and that the sugar levels of both Riesling and Cabernet Sauvignon have been rising over time.
Average Attribute Values per Grape Type
Similarly, customer marketers should conduct the same type of analysis for each customer segment, gaining key insights into each segment’s identity, based on its various attributes. In this way, one can begin to see each customer segment for what it really is.
The following example from Optimove’s Segmentation Explorer dashboard presents the values given to seven different customer attributes: days since registration, longevity, past value, future value, customer lifetime value (LTV), days since last order, and net revenue.
Using this dashboard, we can examine key attributes of different customer segments according to their various geographic regions.
We can then drill down further to examine the current values and over-time trends of each attribute across the geographic segments. For instance, in the chart above we can see that Canadian customers currently have the highest LTV in the last period, and it has been increasing for the past three periods.
When examining different segments across different attributes, we can gain insight about which trends exist in the customer base, as well as why they exist.
Step 3: Comparing Cohorts Over Time
Let’s return once again to the vineyard. Pierre now wants to measure how the quality of the wines he produces changes over time. For this, he uses the concept of “vintage wines” (vintage 2006 wines, for example, are wines made from grapes harvested in 2006). Vintage 2006 wines could reach their peak in, perhaps, 2014. Although they may have behaviors and attributes that change over time, wines created from a 2006 harvest of a certain grape will forever be vintage 2006 wines. With vintage wines, Pierre can determine the quality of each wine type and the different accelerations for the time of peak quality for each. Riesling, for instance, may peak after three years and then deteriorate, whereas Cabernet may peak more quickly before deteriorating.
Similar to vintage wines in our vineyard analogy, there is a concept in customer data analysis called cohort analysis. A cohort analysis isolates different customer segments, tracking and comparing them over time, in order to highlight the differences between them. In the example below, we monitor cohorts, tracking their survival rate and the cumulative net revenue of customers with high, low, and no return rates.
We can see in the cohort analysis above that both customers with high and low return ratios have higher cumulative net revenues than customers with no returns. These customers also have higher survival rates. This customer behavior may be counterintuitive, as one might expect customers with a high return ratio to bring in less revenue. Once we follow a specific segment of customers over time, however, we gain a degree of insight into their behavior that we wouldn’t otherwise be able to, which is the perfect foundation for building a successful retention strategy for that segment!
Derive Deeper Insight from Your Customer Segments
A deep understanding of a business’ customer segmentation contributes greatly to the way marketers can manage relationships with their customers. The three-step process described here can help marketers better understand their customer segments, as well as derive deeper insight from them.
Step one is to measure the size of each segment, looking for any noteworthy growth trends. Step two is to measure the current levels and trends of the various attributes of each segment, gaining a deeper understanding of their internal components. The third step is to compare between the different customer segments over time, seeking to highlight the differences between them, and understanding what makes each one’s behavior unique.
The resulting insight serves as a strong foundation for smarter customer marketing strategies, a foundation that moves marketers much closer to maximizing customer engagement and lifetime value.
Pini co-founded Optimove in 2012 and has led the company, as its CEO, since its inception. With two decades of experience in analytics-driven customer marketing, business consulting and sales, he is the driving force behind Optimove. His passion for innovative and empowering technologies is what keeps Optimove ahead of the curve. He holds an MSc in Industrial Engineering and Management from Tel Aviv University.
