“The beginning of wisdom is the definition of terms.”

-Socrates

What’s in a name

We group everything. We like to group our personalities into buckets. We group animals into kingdoms, businesses into industries, and even our bodies into types. While we are not assigning a permanent, single-dimensional, view to whatever we are defining, we are organizing things to better help us understand. So, in an effort to gain wisdom, we simply define things by grouping them together.

Let’s translate this into business, specifically in how it relates to our customer or followers. 

Not all customers are created equal. They each have different needs, wants, and desires and each interact differently with your brand. Some are casual interactors, some might be net promoters, while others are still feeling out your services. We can’t and shouldn’t treat everyone the same and expect the same response from everyone. Instead of treating our customers the way we would like to be treated, we should treat them as they would like to be treated.

Tweet Me- A Twitter Example

For this project we will look at a specific Twitter account of a nationally known brand and try to organize their followers. We could realistically use any data from any data source but Twitter is a fun one to look at.

Describe a User

First task is to get info on each user. This can be anything from personal demographic info (age, gender, zip code, etc.), to info about how they interact with the company brand (number of transactions, average purchases, etc.). You just want to make sure the chosen fields are applicable for each person. For this project we pulled ~24,000 follower’s data.

For our brand on Twitter we will pull the following for each follower. Twitter provides all of this:

• Id- The ID which is unique to each user
• Created Date- the date and time their account was created
• Favorites Count- the number of favorites a user has given
• Followers Count- the number of followers a user has
• Friends Count- the number of friends a user has 
• Statuses Count- the number of status the user has
• Profile Background Color- the color of the user’s background on their twitter account (some people are all about customization)
• Profile Text Color- the color of the text of the user uses on their twitter account 
• Verified- if the user’s identity has been verified by Twitter

Some of this information may seem not relevant or needed. However, since it explains something about each person it provides dimensionality to help us get to know the user better. 

Someone who changes their profile background and has a large following is very different from someone who is verified and has a small following. If we could get even more information on each individual we could try to do so.

Infer Information

Using the info above that Twitter gives us we can infer a couple of things. 

First, we can calculate the number of days each person has been on the social media platform by subtracting today’s date from the date created. 

Second, we can guess the default color of text and background by looking at the most frequent color for each user (see color samples below). So, if the user changes either the text or the background colors from the default, we know they are pretty serious about their Twitter game. (Fun Fact: I didn’t know you could change either of those until this project!)

Grouping

The fun part now begins because we can give the computer this information and we can help it find hidden trends and groupings we can’t see on our own. Below is a visual of all 24,000 followers (dots) and the groups they belong to (colors).

The groups may look really close together but it’s just for visualization purposes; they are very distinct. 

Let’s regroup and see what we have real quick:

1. We have a 5 distinct groups for all the followers 
2. We know which group each follower is in
3. We also have the template which gave us these results. So in the future, if there is a new follower, we use this template to place them into one of the groups.

Defining the Groups

Each groups size is listed in the table to the right.

Group	Number of Followers
1	6,233
2	3,700
3	4,312
4	6,098
5	3,857

Looking at each group individually we can start to infer some things about them. We aggregated the number of followers, number of verified users, average number of days on Twitter, etc. and were able to find some interesting trends. 

This part is the most fun for me. Here we have the information from the grouping, but it takes an expert on the data, the business, to understand the groups. This is where art meets the science and perfectly shows how data is just a tool (albeit a very powerful one) for a business owner.

Here are the groups based upon my understanding of the Twitter business and the aggregation totals. I even gave them some names to help remember them:

1. New Kids on the Block– Group 1. They are young on twitter, less than 2 years, not many followers or favorites, but they do customize and some of them are verified. 

2. The Influencers– Group 2 they have high numbers in everything they do. They customize their twitter, have the most verified accounts, most followers, most status’, and have been on twitter the longest. They also have almost the same amount of status as they do favorites. 

3. Younglings– Groups 3. They have been on Twitter the same amount of time but neither of them are doing too much. Having a few more favorites, they are scrolling and liking things. But they don’t have anyone who is verified.

4. Recreationals– Group 4 is doing everything, but they’ve been on Twitter the second longest. These people seem savvy at Twitter, they have friends and followers and they have status’ as well. But they aren’t verified and don’t spend all their time on Twitter, at least not as much as the Influencers who are working hard in every category.

5. The Alchemists- Group 5 is the only group that has less favorites than followers, friends or status’. They also are the only group that has more followers than status. They aren’t verified, they don’t customize much compared to the other groups but there is something about what they are posting that makes people want to follow them.

Now What?

Now that the groups are defined and named we can get to know each group uniquely. We can start looking at individual users in each group and get a feel for what they like, what they post about and what they follow.  

After our research activity we might change the names around a little or improve on our above descriptions.

Then we can give them what they want. Assuming we are the brand, we can then segment our marketing or even products based on these groups. 

When we develop new products and want to know what our customers want we can instead focus on these groups instead of looking at the aggregate. 

These groups also help us if we wanted to get the word out about a new product or offering. To do so, we could look at what the Influencers and Alchemists are posting about and cater our message to them. 

Customers are already telling us what they like by how they interact with us, we just have to listen.

Additionally, as mentioned before, we saved this template. Now, any new follower we get, we can quickly pass them through the template to assign them to a group. This allows us to better understand them and actually provide them custom, curated content immediately.

Final Thoughts

With a simple grouping exercise we now have an added lens to help us serve our customers better because we know who they are and have a better idea on how to serve them. We can now treat them the way they would like to be treated.

Just like anything with data, grouping does not provide all the answers to our business problems but does give greater clarity to help make those decisions easier and faster.

Key Takeaways

1. Data points can be gathered and then grouped together in a dynamic way for definition and understanding.
2. Once grouped, research into each group can yield profiles with great depth.
3. Specific, curated, content/products/offerings can then be created for each group.
4. When new data points come, they can quickly be grouped and welcomed in and presented with content already developed for the group. Needs can also be quickly anticipated.