Homework answers / question archive / TiVo Segmentation Analytics John McCambridge, recent business school graduate and newly hired analyst at ThinkAlike, a boutique marketing consulting firm, was feeling stressed about his first project

TiVo Segmentation Analytics John McCambridge, recent business school graduate and newly hired analyst at ThinkAlike, a boutique marketing consulting firm, was feeling stressed about his first project

Computer Science

Share With

---

TiVo Segmentation Analytics

John McCambridge, recent business school graduate and newly hired analyst at ThinkAlike,

a boutique marketing consulting firm, was feeling stressed about his first project. It was October

24, 1998, and four days earlier, Leslie Pantelle, head of marketing for TiVo, had visited the firm’s offices unexpectedly.

“We’re about to launch a new product—a groundbreaking device like nothing you’ve seen before,” she said. “We’re going to revolutionize the way people interact with their TVs. It’s going to

be huge! The market is ready, our investors are excited, and the product is phenomenal.” Pantelle

continued, “Our biggest challenge at this point is segmenting the market. We’ve come up with

some ideas internally but we want an outside point of view.”

The partner in charge had agreed to take on the project even though TiVo set an aggressive

deadline of one week. Since Pantelle’s visit, ThinkAlike had gathered some robust data for TiVo.

The presentation was scheduled in three days, and it was McCambridge’s task to analyze the

survey and draft segmentation recommendations for the partner managing the project.

Television Viewing in the United States

Americans had a love/hate relationship with television viewing. Of the country’s 100 million

households, 98 million owned at least one TV. The average household had 2.4 TVs and spent

seven hours and fifteen minutes per day watching television. Traditional broadcast channels were

supplemented by other services: in 1999, 78 million households spent $34.4 billion on cable TV

service, and another 14.5 million households paid for satellite TV service.

Although Americans spent a great deal of time in front of their televisions, they were far from

satisfied with the experience. Viewers frequently complained that their favorite programs were not

on when they had time to watch TV, and the watching they did was often interrupted by phone

For the exclusive use of M. Ramsaran, 2022.

This document is authorized for use only by Maverick Ramsaran in 2022.T 2

i V o S e g m e n T a T i o n a n a l y T i c S KEL932

K e l l o g g S c h o o l o f m a n a g e m e n T

calls and family demands. The VCR, which might have been used to address these problems by

recording shows to watch later, was difficult for many viewers to program.

TiVo’s Origins

In August 1997, Jim Barton and Mike Ramsay founded TiVo in San Jose, California, with the

goal of increasing consumers’ control over their television viewing. They envisioned a system,

which they named TiVo, that would enable a user to search a complete guide of television programs

and digitally record and store selected programs in a set-top box that was referred to as a personal

video recorder (PVR) or a digital video recorder (DVR). TiVo offered its customers numerous

unique features, including:

• Pause, slow motion, and instant replay for live television programs

• Simultaneous playback and record

• Option to skip commercials

• Direct (nonlinear) access to all recorded programs

• Search for and record shows based on a favorite actor, director, genre, or show

• Automatically record a favorite program every time it aired, regardless of time

• Recommendations based on viewer preferences (similar to Amazon recommendations)

TiVo planned to charge a monthly fee of $9.95 for the subscription service that would enable

users to view special programming, receive personalized viewing suggestions, and control

live television. Alternatively, users could purchase a lifetime subscription for $199. Initially the

subscription fee would be the company’s primary source of revenue, but once a critical mass of

consumers adopted TiVo, the detailed data regarding viewer behavior and preferences captured

by the subscription service was anticipated to be a major revenue source. This data could be used

to sell targeted access to advertisers and programmers, as well as to create interactive television.

Barton and Ramsay recognized that they lacked the resources to manufacture the hardware

and build a PVR/DVR brand, so they partnered with consumer electronics leaders Sony and

Philips. The companies agreed to manufacture the set-top boxes, which carried the Sony or Philips

brand along with the TiVo name. TiVo subsidized production costs to ensure the price for lowerend models was under $500.

Barton and Ramsay’s business model was very appealing to investors and they quickly

attracted significant financial backing from prominent venture capitalists. AOL/Time Warner, a

major cable television company, also invested more than $240 million.

How TiVo Worked

When TiVo recorded a television program, the incoming analog video was converted to digital

video and then encoded and compressed (see Figure 1). The compressed recordings were stored

on the hard drive as separate files in order to allow random, or nonlinear, access to individual

programs.

For the exclusive use of M. Ramsaran, 2022.

This document is authorized for use only by Maverick Ramsaran in 2022.3

KEL932 T i V o S e g m e n T a T i o n a n a l y T i c S

K e l l o g g S c h o o l o f m a n a g e m e n T

TiVo’s most important feature was its hard

drive. For cost and compatibility reasons, it

used the same hard drives that were used in

desktop computers. The set-top boxes were sold

in configurations that stored 20, 30, or 60 hours

of basic-quality video, but users could set the

devices to record at higher-quality levels, which

reduced the advertised storage by up to twothirds. The 20-hour model was priced at $299, and

the 60-hour model was priced at $699 (MSRP).

The TiVo set-top box used an existing

telephone landline to download updated

programming information and unique content

every day, usually very early in the morning.

Software upgrades and enhancements also were

automatically delivered via the phone line, but

TiVo was designed not to interrupt incoming or

outgoing calls.

Survey

After Pantelle’s visit, ThinkAlike immediately designed and implemented a survey to gather

its own data about potential users of TiVo. One thousand participants from across the United

States were surveyed by phone for thirty minutes each. The survey asked participants about their

purchasing behavior and their attitude toward the TiVo concept. The sample included a wide range

of ages, locations in the United States, and interests and attitudes toward the product. The survey

respondents were quizzed on fourteen attributes, which are summarized in Exhibit 1.

Analytics and Segmentation

Segmentation is an art, but it relies on rigorous mathematics known as principal component

analysis (or “factor analysis,” a simplified version of it). The essence of principal component

analysis is identifying attributes in the data that are correlated with others, and then removing

them so that the remaining attributes separate the population into homogeneous segments.

ThinkAlike created a table listing all fourteen attributes in the rows and columns to record the

correlation of attributes—cells were filled with the correlation of each attribute with every other

attribute (see sample table in Exhibit 2). The value of every cell on the diagonal was 1 because

every attribute was perfectly correlated with itself.

Attributes with non-numeric responses needed to be coded by assigning a number to each

response; for example, Gender could be coded by assigning the value 1 to every male, and 2 to

every female. Once all non-numeric responses were coded the attributes could be correlated.

If Gender and Marital Status had a correlation (or r2) of 0.92, as shown in the sample in Exhibit

2, then Gender provided a high indication of Marital Status. This strong correlation could occur

Figure 1: TiVo Video Processing

Source: Alice M. Tybout and Julie Hennessy, “TiVo,” Case

#5-104-024 (Kellogg School of Management, revised

September 14, 2011).

For the exclusive use of M. Ramsaran, 2022.

This document is authorized for use only by Maverick Ramsaran in 2022.T 4

i V o S e g m e n T a T i o n a n a l y T i c S KEL932

K e l l o g g S c h o o l o f m a n a g e m e n T

if nearly all male respondents were single and nearly all female respondents were married, for

example. Because it was highly correlated with another attribute, Gender could be ignored for

purposes of segmentation.

Running a correlation between all pairs of attributes would identify which of them were highly

correlated and which were not. (A correlation or r2 above 0.45 is regarded as high.) In the example

in Exhibit 2, Location and Annual Income were highly correlated with at least one other attribute,

so they also could be ignored for segmentation purposes. The remaining three attributes, which

were not highly correlated with anything but themselves—Marital Status, Education, and Age—

would be the basis for three distinct segments.

Analyzing the Survey Data

McCambridge assumed TiVo’s target segment would have similar attitudes toward the

product, but that would not be enough to define a useful segment. What if people who liked the

product could not afford it or were unable to use it in a way that satisfied them? What if they were

interested but were not early adopters?

For now, he had to start working with the data