In the first part of our series on customer segmentation for retail, we will explore the differences between using what a customer looks like on paper and using what they do to inform marketing strategies.
In this series, we're going to get into quite a few different techniques and best practices for discovering segments. Then we’ll present a few common challenges that e-commerce retailers face and how segmentation can help solve those problems.
But first, why do we segment?
Before we get into the nuances of segmentation, it’s important to set out its purpose. Otherwise, it’s all too easy to be led astray by flashy tools and tips and tricks and hacks.
The goal of any form of segmentation should be to leverage a better understanding of our customers so that we can improve acquisition and retention. That’s all.
What is customer segmentation?
Every retailer knows that not all customers are the same. It's a basic principle. If we can bring to life how different types of customers are different, it can really change the way that we market to and connect with our customers.
What are some common ways that e-commerce companies are segmenting their customers?
The oldest form of segmentation—because it is often the easiest form of segmentation—is demographic segmentation. This kind of segmentation describes and groups customers by various hard, concrete attributes, such as:
This used to be contrasted with behavioral segmentation, which focuses more on the actions that users have taken. Some common forms of behavioral segmentation here are:
- What was the first thing that somebody bought?
- How often does someone buy?
- What types of products do they buy?
- What brands does someone buy?
- What device types did they use to make purchases?
- Are they a new customer, an older customer?
- What Lifecycle Stage are they in?
We can use Behavioral and Demographic together, but they’re very different forms of segmentation. The big thing that anyone defining personas or segments needs to come to terms with is that actions speak louder than your age, gender, skin color, income, or any other demographic attribute.
You can imagine that you and your across-the-hall neighbor are the same age, gender, and race. You have the same job. You make the same amount of money. But you like entirely different things, and you buy different things in different patterns and cadences.
That's why behavioral segmentation is so important.
Wait, remind me. Why do we segment?
To improve customer acquisition (i.e., to find more and better customers) and to improve customer retention.
Regarding Acquisition, you can imagine if you can find the type of customer that resonates most with your business and brand, you would want to find more of them. Then maybe you find that this type of customer primarily converts through Facebook. That information will (or should) then dramatically affect how you tailor your Facebook campaigns.
As you go from channel to channel to channel, you might discover meaningful segments associated with each one. This information, used properly, can really improve your chances to connect with potential customers and to convert those customers.
Retention is a little bit more straightforward. When you know the types of things that your customers want, you can send them relevant information about those things. When you know the ways in which this valuable segment behaves differently from this other valuable segment, you can communicate with them differently for maximum effect.
The cycle you want to encourage goes like this:
When you present stuff that customers care about >>>
The customers are more likely to buy it >>>
And more repeat purchases build a happier business >>>
Then you can expand your inventory >>>
To include more stuff that people care about >>>
And your customers are more likely to buy it >>>
And more repeat purchases build a happier business >>>
And the happy cycle continues.
A Tale of Two Segments
Here’s an example to illustrate ineffective versus effective segmentation.
Let's assume you’re a fashion apparel company and you sell pants, shorts, shirts, and all kinds of goods.
You have this hypothesis that maybe women are more interested in buying shirts and men are more interested in buying pants.
You then segment your customers into the Men Group and the Women Group. You send an email to the men about the new pants in season and likewise, send an email about the hottest new shirts that you have in stock to the women.
The problem is that it isn't necessarily the case that men only like pants nor is it necessarily the case that women only like shirts. In other words, your hypothesis can be wrong and you miss out on selling to pants-loving women and shirt-loving men.
You chose to segment your customer base into men and women because you thought that certain product preferences were associated with those attributes, but it's actually not true.
That's an example of a demographic segmentation that might not work out.
On the other hand, let’s consider a simple form of behavioral segmentation, like first purchase.
If the first thing the customer bought was pants, we'll send them more information about our pants. If their first purchase with us was a shirt, we’ll send them emails about our shirts. Because this is behavioral segmentation, it’s extremely likely that there will be men and women in both categories.
In this oversimplified example, by segmenting based on customer behavior, you’re more likely to get a better response rate.
This is a very simple world. We are assuming, if you bought pants once, you want to buy pants again and if you bought shirts once, you want to buy shirts again. When we move beyond the static segmentation of gender or the place where you live, we often see a tremendous lift in response rates even from something so simplified as this.
Of course, it can get better and better than this as well.
The key thing about segmentation is that it should tell a meaningful and relevant story about your customers.
If you're selling socks, it might not make sense to break your population into those who are passionate about birds and those who aren't. It might not make sense to segment your customers into people who have red hair and people who have blonde hair.
Demographic segmentation may help you filter things if you have gender-specific socks, but behavioral segmentation based on things like purchase cadence and product affinity will tell a more meaningful (and this useful!) story.