Looker Blog : Data Matters

Creating Actionable Customer Segmentation Models

Dan LeBlanc, CEO & Founder of Daasity

Jun 27, 2019

What is customer segmentation?

Customer segmentation is a way to split customers into groups based on certain characteristics that those customers share. All customers share the common need of your product or service, but beyond that, there are distinct demographic differences (i.e., age, gender) and they tend to have additional socio-economic, lifestyle, or other behavioral differences that can be useful to the organization.

What type of information is used in customer segmentation

Any information you can acquire about individuals can be used to create a customer segmentation. Direct-to-consumer brands and B2B companies are at a distinct advantage because of the amount of information they can obtain about their customers just from their transaction data alone.

Basic data types typically include:

  • Geography (billing info, shipping info (if applicable), browser info)
  • Product(s)/Service(s) purchased
  • How customers found you (referring URL and/or campaign info, promo codes)
  • Device used (device type, brand (if mobile), browser)
  • If this is a customer’s first purchase
  • Payment method

Beyond these basics, companies may choose to collect more information as part of the selling or checkout process that can augment their customer data, such as:

  • Reason for purchase
  • Marketing or advertising channel that drove purchase*
  • Intended use: business, personal, self-consumption, gift, etc.
  • Company industry segment
  • Job title
  • Age/Gender

*Important Note:
This has become more common, especially with direct-to-consumer businesses trying to assess their marketing efficacy and offer another viewpoint besides last-click in Google Analytics. There is always a healthy margin of error applied to data reported in this way from a customer, but it certainly indicates what they believe to be the most memorable or important reason for their purchase. Daasity has built out specific logic for processing this information, along with other data, to help determine the most likely marketing channel responsible for purchases.

From here, there is the opportunity to either infer additional attributes or purchase additional attributes. Inferring attributes means you have already collected data that results in a strong correlation to another attribute. For example, you might infer gender from name.

The other option is to purchase data and append it to your customers’ existing profile data. Companies like Experian, Acxiom, and others happen to have significant amounts of purchase data from credit card transactions, as well as demographic data that they have mapped to certain behaviors. They have strong match rates to provide additional data, (referred to as 3rd party data) such as:

  • Estimated household income
  • Presence of children
  • Homeownership
  • Amount of spend in your company category or other retail categories
  • Lifestyle or behavioral interests

6 types of customer segmentation models

Common customer segmentation models range from simple to very complex and can be used for a variety of business reasons. Common segmentations include:

  1. Demographic
    At a bare minimum, many companies identify gender to create and deliver content based on that customer segment. Similarly, parental status is another important segment and can be derived from purchase details, asking more information from customers, or acquiring the data from a 3rd party.

  2. Recency, Frequency, Monetary (RFM)
    RFM is a method used often in the direct mail segmentation space where you identify customers based on the recency of their last purchase, the total number of purchases they have made (frequency) and the amount they have spent (monetary). This is often used to identify your High-Value Customers (HVCs).

  3. High-Value Customer (HVCs)
    Based on an RFM segmentation, any business, regardless of sector or industry, will want to know more about where HVCs come from and what characteristics they share so you can acquire more of them.

  4. Customer Status
    At a minimum, most companies will bucket customers into active and lapsed, which indicates when the last time a customer made a purchase or engaged with you. Typical non-luxury products consider active customers to be those who have purchased within the most recent 12 months. Lapsed customers would those who have not made a purchase in the last 12 months. Customers may be bucketed even further based on the time period in that status, or other characteristics.

  5. Behavioral
    Past observed behaviors can be indicative of future actions, such as purchasing for certain occasions or events, purchasing from certain brands, or significant life events like moving, getting married, or having a baby. It’s also important to consider the reasons a customer purchases your product/service and how those reasons could change throughout the year(s) as their needs change.

  6. Psychographic
    Psychographic customer segmentation tends to involve softer measures such as attitudes, beliefs, or even personality traits. For example, survey questions that probe how much someone agrees or disagrees with a statement are typically seeking to classify their attitudes or perspectives towards certain beliefs that are important to your brand.

5 Benefits of customer segmentation

There are several benefits of implementing customer segmentation including informing marketing strategy, promotional strategy, product development, budget management, and delivering relevant content to your customers or prospective customers. Let’s look at each of the benefits in a bit more depth.

  1. Marketing Strategy
    Customer segmentation can help inform your overall marketing strategy and messaging. As you learn the attributes of your best customers, how they are alike, and what is important to them, you can leverage that information in messaging, creative development, and channel selection.

  2. Promotion Strategy
    An overall promotion strategy (i.e., our customers are deal seekers, therefore we should offer frequent deals) for sending promotions for specific segments can be made better with information from a broad customer segmentation scheme. You may find that certain cohorts of customers don’t require discounts when you use certain messaging, thereby saving you from having to offer a discount for those groups at all.

  3. Budget Efficiency
    Most companies do not have unlimited marketing budgets, so being precise about how and where you spend is important. You could, as an example, target similar customers to segments of high value or those most likely to convert to get the most return from your marketing investment.

  4. Product Development
    The more customers you acquire, the more you learn about what is important to them, what features they want, and which customers are the most valuable. Your company can use these insights to prioritize product features that either appeal to the most customers, those categorized as high-value customers, or other characteristics that makes sense for your industry.

  5. Customers Demand Relevance
    Whether it’s D2C, B2B, Millennials or GenZ; it seems that there is a study or resource on every possible group of customers stating that relevant content is important to them. These customer segments are more likely to respond, buy, and respect the brand and feel connected if provided with relevant content. By performing some level of segmentation, you can ensure that the messages you are delivering via email, on site, through digital ads, or other methods are targeted and relevant to the individual seeing it. It is almost counter-intuitive to the hyper vigilance of data privacy to use so many pieces of data in this way, but with so many marketing messages coming at people today, no one has time for something that isn’t relevant to them.

How to make customer segmentation actionable

To make your customer segmentation actionable, first, you must start with a goal in mind. As previously mentioned, segmentation can be simple, complex, or anything in between — and you aren’t limited to one set of segments. With the ease and accessibility of data today, you can devise different customer segments for different purposes.

The amount of information that can be obtained from various sources is endless. But, it’s only useful if you can use it. This requires questioning, being curious, and analyzing the data you have. From there, as you find treasures buried in the data you have, design a test to confirm that is in fact a useful finding.

Examples of customer segmentation

Target has perhaps the most famous story of using customer segmentation, analytics, and marketing techniques to increase their share of wallet with pregnant women. In 2012, the incredible story broke of Target accidentally informing a young woman’s father that she was, in fact, pregnant, before she had broken the news to him herself.

Once a customer has a child, his or her purchase patterns and basket contents suddenly change to contain diapers and other products consistently. That is a whole segment of customers: people who’ve just had babies. Add gender to it and you have women who have just had babies. As the analysts evaluated this segment’s history, they started to see purchase patterns emerge as markers of the pregnancy’s milestones. From here, they surely built predictive models that would classify customers as they hit some of these markers and flagged those customers as newly pregnant. The action that Target took was to market very specifically to these women with highly targeted ads and direct mail for baby items, baby clothes, and supplies. When a young woman received one of the mailers addressed to her, her father was astonished at how foolish and careless Target would be...until he found out that his daughter was indeed expecting, and Target knew before him.

This example is extreme but memorable. Segmentation can be employed using knowledge of your customers, knowledge of your business, common sense and perhaps a few creative variations — even if you don’t have a Target-sized team of data scientists pouring through the data.

An easy way to use segmentation and to start collecting data for immediate results is through email campaigns. Let’s say you are planning a campaign series and really want to learn how different customer groups react to various messaging and offers. You have a healthy database of emails that includes a mixture of customers and non-customers. Using the code below, you can group customers into non-customers, and then groups based on recency of last purchase being 0-3 months, 3-6 months, 6-12 months, and >12 months.

view: customer_recency {
  derived_table: {
    sql:
  WITH
    last_order AS
  (
    SELECT
      customer_id,
      MAX(order_date) AS last_order_date
    FROM order
    GROUP BY
      customer_id
  )
  SELECT
    c.customer_id,
    CASE
      WHEN DATEDIFF(day, CAST(lo.last_order_date AS DATE), CAST(current_timestamp::timestamp AS DATE))  BETWEEN 0 AND 90 THEN '1: 0-3 Months Active'
      WHEN DATEDIFF(day, CAST(lo.last_order_date AS DATE), CAST(current_timestamp::timestamp AS DATE))  BETWEEN 91 AND 180 THEN '2: 3-6 Months Active'
      WHEN DATEDIFF(day, CAST(lo.last_order_date AS DATE), CAST(current_timestamp::timestamp AS DATE))  BETWEEN 181 AND 365 THEN '3: 6-12 Months Active'
      WHEN DATEDIFF(day, CAST(lo.last_order_date AS DATE), CAST(current_timestamp::timestamp AS DATE))  > 365 THEN '4: 12+ Months Lapsed'
      ELSE 'Non-Customer'
    END AS customer_recency_group
  FROM customer c
  LEFT JOIN last_order lo ON c.customer_id = lo.customer_id
  GROUP BY
    c.customer_id,
    lo.last_order_date
      ;;
  }
  dimension: customer_id {
    sql: ${TABLE}.customer_id ;;
    primary_key: yes
  }

  dimension: customer_recency_group {
    type: string
    sql: ${TABLE}.customer_recency_group ;;
  }

  measure: num_customer {
    type: count
  }
}

daasity

You can then evaluate the performance of each group against sent content to determine if there are specific messages that resonate more.

Accomplish more with actionable customer segmentation models

Customer segmentation is an important part of any business aiming to grow revenues, repeat rates, share of wallet and profitability. Segmentation does not have to be incredibly complex or expensive, and it can be easily accomplished using a Looker dashboard with readily available transaction or demographic data. And customer segmentation benefits your customers and your organization, allowing your customers to feel more connected to your brand because they’ve been received relevant content, and in turn your company should see increased positive results.

Subscribe to the Looker blog below to stay up to date on trending topics in big data, data analytics stories, product news, and more pieces from our Daasity team.

Previous

Subscribe for the Latest Posts