Customer Segmentation for Growth Hacking
Marketers often equate growth hacking with driving user acquisition.
They invest all their effort and money to acquire maximum users with a minimum amount of spend. However, not all users are equal. Some will come on your website, create a profile, and never buy a thing from you. Others would consume everything on your website and also purchase every new item you offer. And the rest will fall somewhere in this spectrum. And wouldn’t you be willing to pay a lot more to acquire the latter kind of customers? And thus wouldn’t you want to know the different types of customers to be able to make such trade-offs?
Let’s look at this in more detail with an example. Let’s say you own a women apparel subscription business — EverYoung (EY). EY sends one set of branded matched-up work attire every month for $39.99/month to its subscribed customers. EY’s value prop is guaranteed satisfaction and the best value for money. EY achieves this by asking the customer for their preferences using their style survey. The survey identifies preferences on color, style, fit, body type, size, age, skin tone, etc. using a series of well thought out curated questions. Then EY creates a custom box of work attire site for every customer, every month. There is never a repetition of clothes for any customer. And customer support is ever-present to address any size/fit issue.
EY’s marketing engine delivers leads to EY’s website where the user is prompted to create their custom style-profile tied to their Social account to get a sneak preview of a sample box, created just for them. A percentage of those who create the profile, then go on to subscribe for the monthly box. EY marketing delivers these leads from various sources, including Pinterest, Instagram, Facebook, Google, blogs, and other partner sites. Also paid advertising on Google and bing deliver leads too.
To date, EY treats all leads equally, and every individual goes through the same experience of answering the style survey and then being shown a delightful sample box and then being offered a promotional deal to sign-up.
Now imagine, if you could know, based on the style survey of the customer and their social profile — whether this person is price sensitive or not? Wouldn’t you show them different promotional deals based on that information? Consider if you could know whether the person is likely a long term customer or short term, wouldn’t you be willing to spend more or less based on that information? What if you knew — that this profile prefers a greater variety of colors and style vs. uniformity of styles and color? What if knew- this profile was the adventurous kind and would love to see a pattern they have never seen before.
You get the idea. If you could segment your prospect base, based on all their information post the style survey, you could customize your offer and message. Thus, you would not only engage with them more meaningfully at the point of offering them the subscription but also in future engagements towards improving your conversion. You could also drive better lead conversion before they are delivered to your site by customizing the offer and messaging based on channel. This customer segmentation approach would drive significant user growth, and not 4–5% but 400–500% more paid subscribers. And who doesn’t want that?
So, how do you achieve this?
Let’s start with a simple segmentation first. That’s how we start with our clients as well. Simple segmentation is achieved by you deciding 2–3 meaningful dimensions to segment your customer base. We then use those dimensions to segment your entire past user base — customer as well as prospect base. For example, you may choose the color, style, and age to segment your base. Let’s say you aggregate the colors into three categories
1. BWG: black-white-grey
2. Pastels
3. Deep
You similarly categorize the style into — bold, varied, and subdued; and age into millennial, mid, and older.
If you segmented your base on these three dimensions, each with three levels, you would get 27 segments, for example.
1. Millennial — BWG — bold
2. Older — Deep — Varied
And so, on. Now, not each of these segments may have enough count. For example, there may be very few Millennial-deep-bold and millennial-pastels-bold, which means you can combine all millennials-bold into one segment. Let’s say you end up with 15 sizable segments after all this collation.
Now you would want to profile these segments on various dimensions on which you want to take action. For example, you may find that majority of Millenials-bold are acquired from Pinterest, and their average conversion from survey to a paid subscription is 2% (your average conversion is 1%), i.e., your millennial-bold convert more. Your messaging and promotional offer is working better for them than other segments. You can then start using this information to create your channel-segment level customization to improve conversion. Initially, you would have lots of hypotheses on improving conversion. As you develop a greater understanding of your segments and you A/B test UX-offers-messaging combinations, you will start creating above-average customized experiences and serving every segment with what they need and want.
However, as you can imagine, these segments you have developed have only considered 2 or 3 dimensions. You are collecting much richer information from each customer. You know where they are coming from and their social profile. Given that you have PII on your customer, you also have an option of external data append to add information like income, etc. to the mix. With this richer set of dimensions, you can create even better segmentation where you can group customers who think and act alike and have similar needs into one category. And this is where we start getting into segmentation using advanced machine learning or AI algorithms. With this, you may be able to create a distinct persona like Stylish Cindy, Adventurous Ayla, Value-seeker Vilot, Basics-Bianca, and so on. To learn more about advanced machine learning methods, you can download our case study of customer segmentation using K-Means Clustering here.
All the best and happy growth hacking!
