Augmented Intelligence
Jan 03, 2017

A second look at long tail

"Long tail" describes the phenomenon in data distribution when there are a large number of data points far from the central part of the distribution. The heavy part is called the "head" and the thin part is called the “tail." Long tail is complementary to the power law: they are two sides of the same story. According to the power law, the majority of incidences come from a small percentage of sources. For example, the famous 80/20 rule states that 80% of the effects come from 20% of the causes. The remaining 80% of the items, which account for 20% of the occurrences, are called the “long tail."

Long-tailed distribution is exhibited by diverse phenomena—e.g., income, popularity of words, etc.—although the split may not always be exactly 80/20. In business, sales data distribution often has a long tail: a small percentage of customers contribute to a large proportion of sales, while customers located in the tail area do not purchase frequently.

Image of long tail 

While studies about the long tail started as early as the 1940s, the concept was popularized by Chris Anderson in 2004. While the power law is discussed in almost all industries, long tail is mainly discussed in insurance, finance, and retail. Since the implication from the power law is to focus on your customers who generate the most revenue, targeting customers with the highest market value has been the dominant force in the development of marketing strategies. Traditional KPIs look good by taking care of the most influential customers—even when this cohort is only one-fifth of your customer base. It seems justifiable to leave out the bulk of low-revenue customers and focus on customers with the largest revenue. Before we rush to agreement, let's ponder over a few points which support the opposite view:

  • Customer centricity: Marketing has shifted gradually from product-centricity to customer-centricity. Today it is more crucial than ever to take customer needs into consideration and enhance customer experience. Directing insufficient attention to the majority of customers means shifting from a customer focus to a product focus. But business will suffer in the long run if customers' needs are not properly accounted for
  • Marketing cost: Marketers shy away from targeting customers in the tail region because the return on these customers does not look good. In traditional marketing, where sales is the main channel, the cost of marketing is flat across all customers and therefore priority is given to customers with high value. Nowadays, emerging channels make it easier to reach potential customers at a lower cost. The question then becomes how to leverage non-traditional channels to generate the impact for the long-tail customers and deliver the right message at the right times. Successful stories abound in e-commerce and the fashion industry
  • Customer migration: The long tail reflects reality at one point in time. Customers in the tail region do not always stay there. Migration can be both inbound and outbound. Instead of treating long-tail customers as low priority, marketers should proactively explore ways to upgrade them from tail to head. Meanwhile, they should strive to retain high-value customers with the propensity to downgrade into the long-tail region. Treating long-tail customers as a static group is myopic and kills business growth potential

As the number of media/channels for targeting has increased, many marketers have become disoriented: they find themselves lost in the information ocean. However, more choices in targeting enable us to differentiate communication frequency, media combinations, and messaging to meet the needs of customers, regardless of their position on a distribution chart. It is up to marketers to decide how to leverage analytics to shift customer dynamics and achieve business growth. We would like to hear your opinions on the long tail concept and how it affects marketing strategies within your industry. Feel free to throw in your thoughts!

About the author

Jingfen Zhu

Jingfen Zhu

Chief Scientist, Analytics, Chief Science Officer

Follow Jingfen Zhu on LinkedIn