Segmentation Roadmap Sept 5 2017

Prospect Segmentation
Roadmap:
Current & Future Schemas
Chudi Okoye
August 2017
D2C| SEGMENTATION,
TARGETING
T-Mobile Confidential
Acquisition Team’s Mission
D2C mission is to smartly leverage

data to define and target consumer
segments and deliver relevant,
personalized messages to them about
T-Mobile products and services, in a
‘multi-channel’ fashion
D2C| SEGMENTATION,
TARGETING
T-Mobile Confidential & Proprietary Information
T-Mobile Confidential 2 2
Expected Uses of Consumer Segmentation
 Goal: Move beyond mass media targeting (i.e. 18-49 year olds) and one message to all
 Hypothesis: Algorithm-based targeting and differentiated offers/messaging will improve efficiency and drive incrementality
 Clustering/Targeting: Identify, profile and size market clusters, and develop data-driven targeting
strategy
 Offers: Differentiate offers (rate plan/device/promos) for specific targets to drive response and
Primary incrementality
A Marketing
 Messaging/Positioning: Differentiate TMO messaging by aligning product values & RTBs to target’s
Uses
needs
 Funnel and Channel: Differentiate funnel and channel treatments to improve/expedite consumer
journey
 Device: Align T-Mobile consumer profiling with OEM segmentation for consistent product
messaging
Other
B Marketing  Prospect CLV: Align prospect segmentation to CLV model for consistent economic assessment with
base
Uses
 Retail: Use segmentation to enhance local market insights to drive retail location strategy
D2C| SEGMENTATION,
TARGETING
Consumer Segmentation Roadmap: Where We’ve Been and Where We’re Going
 Evolution to a holistic segmentation scheme combining a behavioral algorithm and a Magenta brand affinity
model Op
High tim
i ze
d
4. Finalize with a
Magenta brand
Ca
affinity model
m p
h t
aig
3. Build algorithm-
esig
or
Prospect Specificity
ns
based behavioral F
segmentation
2. Created basic
demographic
segmentation
1. Developed
“Magenta People”
segmentation
s ight
In
Low/Start Delivery Timeframe End
(1) Q3 - Q4 2016 (2) Q4 2016 (3) Q1 2018 (4) Q2 2018

Source: In-house T-Mobile Depiction
D2C| SEGMENTATION,
TARGETING
Where We Were
D2C| SEGMENTATION,
TARGETING
Magenta People Project
Introduction
 The “Magenta People” was a project designed to classify US wireless consumers into broad
segments depicting their status either as current T-Mobile/Metro customers or targetable prospects,
differentiating them from TMUS non-targets
 The segmentation was based on TMUS customer data in Nielsen Mobile Insights and Nielsen
ConneXions
Phase I:
 Created “Magenta People” construct similar to Metro’s “Purple People”
 Identified core ConneXions segments and overlap between Magenta and Purple
Phase II:
 Worked with Metro to reduce overlap by applying additional analytics and business logic based on
demos and wireless behavior
 Re-classified overlapping segments based on modeled affinity to Magenta or Purple
 Profiled and sized new segment groupings and identified additional opportunity segments for
Magenta’s next phase of growth
 Work through implications for media, messaging,
D2C| SEGMENTATION,
TARGETING
creative approach etc.
Model-Optimized Segments for Magenta and Purple People
Axes scores are indices of ConneXions segments derived in analysis of Magenta People and Purple People
“Purple Land” “Common Market”
“New Frontier” “Magenta Land”
Segment Key: Magenta Media Magenta Stretch Magenta Non-Media Purple Secondary Purple Primary Common Market Non-Target
Data Source: Nielsen SMS data, BCMI Analysis/Depiction Bubble area depicts segment market size, scaled to 10%
D2C| SEGMENTATION,
TARGETING
ConneXions Classifications for Magenta and Purple People
ConneXions
Aspirational Magenta Common Purple

Targets People Market People
Purple Primary
Magenta Magenta Magenta Shared Purple
(If Low-
Stretch Primary Secondary Targets Secondary
Scale)
Cyber You & I Techs and Discounts &
Plug & Play Gearing Up
Sophisticates Tunes the City Deals
Gadgets Calling Digital The
Time Shifters Opting Out
Galore Circles Dreamers Unconnected
Smart Multimedia Big City, Landline

Cinemaniacs
Target as Magenta if mid-scale or higher

Gamers Families Small Tech Living
WiFi Video
Technovators IM Nation Last to Adopt
Warriors Homebodies
Broadband Bundled Leisurely
Video Vistas
Boulevards Burbs Adopters
High-Tech The Cyber Techtown
Society Pragmatics Strivers Lites
Generation Antenna New kids on
WiFi Land the Grid
Plugged-in
Families
Target as Purple if low-scale
Kids &
Keyboards
Tech Nests
New
Technorati
Text Legend: Original Purple People SMEs and SMMs
Note: A segment overlapping as Magenta and Purple is assigned as primary target to the cohort with >= 60% share of the segment based on a
logistic model
Consumer status (low- or mid-scale+) determined by: income, credit status, education, employment, number of lines and wireless spend
D2C| SEGMENTATION,
TARGETING
Where We Are
D2C| SEGMENTATION,
TARGETING
Business Context for the Current Ad Hoc Demographic Segmentation
 Purpose
—Creation of a quick segmentation scheme to help drive targeting in 2H 2016
• Recognition that this isn’t long-term plan for segmentation
 Approach
—Leverage sampling of recent activations (last 2 years) to form segments that
reflect the kinds of customers currently being acquired
—Utilize core demographic dimensions from prospect database to facilitate
relevance, speed, and coverage
 Desired outcomes
—Segmentation definition, coloring, and profiling
• 3rd party data - Traditional demographics, summarized credit statistics, etc.
• 1st party data - # lines, MRC, device type, etc.
— Include Nielsen ConneXions codes and Magenta People groupings
D2C| SEGMENTATION,
TARGETING
Demographic Segmentation
 Objective: Get to market as quickly as possible with a “quick and dirty” segmentation, with profiling to drive performance and
learning Final 10 segments
INCOME Groups FAMILY Groups
Segment AGE group I II III IV N Y INDEX TMO customers Penetration Description Count % of US Pop Pen % Shift
1 Millennials 157 53,593 30.8% Millennials < $75K 174,244 13.4% 22.7%
2 Millennials 120 17,067 23.6% Millennials > $75K 72,271 5.6% 12.3%
3 GenX 172 33,911 33.7% GenX < $75K no kids 100,488 7.7% 42.6%
4 GenX 154 43,597 30.2% GenX < $75K with kids 144,456 11.1% 1.5%
5 GenX 97 9,288 19.1% GenX > $75K no kids 48,607 3.7% -3.2%
6 GenX 117 24,037 22.9% GenX > $75K with kids 105,026 8.1% 0.2%
7 Bboom + 39 20,679 7.6% Boomers+ < $75K no kids 271,875 20.9% 2.4%
8 Bboom + 89 24,729 17.4% Boomers+ < $75K with kids 142,005 10.9% -6.4%
9 Bboom + 41 10,689 8.1% Boomers+ > $75K no kids 132,655 10.2% -1.1%
10 Bboom + 83 17,991 16.4% Boomers+ > $75K with kids 109,781 8.4% -9.9%
0 0 0 0 0 0 255,581 19.6% 1,301,408 100.0%
Impact of presence of children GenX versus Millennials Penetration shifts / presence of kids
• As expected, penetration in the younger segments significantly higher than BBoom+ group • Low income GenX is TMO penetrated at a rate equal to or greater than Millennials • Shift away from family plan focus in 2016 from 2015 is reflected in penetration shifts –
• However, within the older group, presence of children shows significantly increased TMO collectively (presence of children doesn’t have an impact) specifically for BBoom+ segments
penetration • Higher income GenX also shows positive impact of children in HH for TMO penetration • Biggest positive shift from 2015  2016 was for the low income GenX no kids segment
Nielsen ConneXions Comparison

Connexion based segments
TMO Data Description
TMO US HH Dist
INDEX2 *** Dist INDEX2 Dist INDEX2 Dist INDEX2
INDEX* Distribution % iOS % > $100 ** % 4+ Lines Magenta Core Magenta Stretch Common Market Purple SMMs
SEG 1 157 13% 41% 55% 73% 16% 119 6% 48 27% 204 20% 152 Millennials < $75K
SEG 2 120 6% 47% 54% 75% 9% 156 9% 168 6% 114 3% 52 Millennials > $75K
SEG 3 172 8% 41% 55% 75% 11% 137 4% 55 13% 174 14% 182 GenX < $75K no kids
SEG 4 154 11% 41% 60% 79% 16% 144 11% 98 25% 227 16% 145 GenX < $75K with kids
SEG 5 97 4% 49% 53% 78% 6% 158 6% 157 2% 62 1% 36 GenX > $75K no kids
SEG 6 117 8% 49% 62% 82% 18% 228 19% 233 4% 50 4% 45 GenX > $75K with kids
SEG 7 39 21% 36% 45% 69% 6% 29 4% 20 6% 26 17% 83 Boomers+ < $75K no kids
SEG 8 89 11% 38% 54% 74% 8% 71 11% 98 10% 90 16% 151 Boomers+ < $75K with kids
SEG 9 41 10% 44% 46% 72% 4% 37 7% 67 2% 18 4% 36 Boomers+ > $75K no kids
SEG 10 83 8% 46% 56% 78% 7% 84 23% 271 4% 50 4% 49 Boomers+ > $75K with kids
100% 100% 100% 100%
Magenta core concentrated in Millennials and GenX HH’s with children (both low and high income) • TMO distribution versus US population distribution
• *** INDEX2 = % of Connexion based segment relative to US population distribution
Magenta stretch concentrated in high income Boomers with children, and high income GenX with children
Purple SMM’s concentrated in low income groups regardless of age or presence of children (Millennials, GenX, Boomers)
D2C| SEGMENTATION,
TARGETING
Demographic Segmentation: Opportunity Prioritization
TMO opportunity: over indexed, but not completely saturated; High
value
Typical TMO targets
Segment 1 Segment 2 Segment 3 Segment 4 Segment 5 Segment 6 Segment 7 Segment 8 Segment 9 Segment 10
GenX < GenX >
Millennials < Millennials > GenX < GenX > Boomers+ < Boomers+ < Boomers+ > Boomers+ >
Definition $75K with $75K with
$75K $75K $75K no kids $75K no kids $75K no kids $75K w/ kids $75K no kids $75K w/ kids
kids kids
Age
TMO 28.9 29.1 43.2 42.6 43.3 43.6 61.9 60.3 61.4 58.9
US Pop
Distribution
29.7 29.7 43.2 42.7 43.3 43.5 68.7 63.8 65.5 61.3
% of US 13% 6% 8% 11% 4% 8% 21% 11% 10% 8%
% TMO (15/16) 21% 7% 13% 17% 4% 9% 8% 10% 4% 7%
INDEX 157 120 172 154 97 117 39 89 41 83
FIC
O
US 657 731 671 664 741 741 696 682 750 750
TMO 619 695 638 636 720 716 661 648 731 725
Distribution (TMO)
Magenta Core 19% 30% 24% 20% 40% 46% 25% 20% 29% 22%
Magenta stretch 5% 21% 6% 9% 25% 30% 10% 17% 34% 46%
Common market 69% 46% 62% 66% 33% 21% 46% 52% 29% 28%
Purple SMMs 7% 3% 9% 6% 3% 3% 19% 11% 8% 4%
# Lines 2.7 2.9 2.9 3.1 3.0 3.4 2.7 2.9 2.8 3.1
iOS 41% 47% 41% 41% 49% 49% 36% 38% 44% 46%
OS
Android 37% 32% 37% 36% 29% 30% 37% 37% 32% 31%
Other 22% 21% 21% 23% 22% 21% 27% 25% 25% 24%
% > $150/month 20.9% 21.4% 21.1% 26.2% 20.0% 25.4% 16.9% 22.4% 15.6% 22.5%
Avg MRC $ $120 $120 $120 $127 $117 $126 $111 $120 $110 $121
CLV per account $966 $1,037 $965 $1,055 $1,106 $1,218 $925 $942 $997 $1,156
D2C| SEGMENTATION,
TARGETING
Usage Example: Linkage to ConneXions and Life-stage Segments
ConneXions By Consumer Segment
Segment 1: Segment 2: Segment 3: Segment 4: Segment 5: Segment 6: Segment 7: Segment 8: Segment 9: Segment 10:
Millennials < Millennials > GenX < $75K GenX < $75K GenX > $75K GenX > $75K Boomers+ < Boomers+ < Boomers+ > Boomers+ >
Code ConneXions Segment LG Lifestage Group Demographic Caption Techno Class Magenta People Classification $75K $75K no kids with kids no kids with kids $75K no kids $75K w/ kids $75K no kids $75K w/ kids Total
1 Technovators F1 Early-Adopting Elite Wealthy Younger w/ Kids High-Tech Magenta Stretch 10.1% 19.0% 2.4% 4.9% 9.2% 40.5% 2.9% 0.5% 8.5% 2.0% 100%
2 Plugged-In Families F1 Early-Adopting Elite Upscale Younger w/ Kids High-Tech Magenta Primary 12.4% 15.9% 3.8% 8.6% 9.0% 36.6% 3.9% 1.0% 7.4% 1.4% 100%
3 Tech Nests F1 Early-Adopting Elite Midscale Younger w/ Kids High-Tech Magenta Primary 31.0% 18.0% 4.2% 15.0% 4.2% 14.9% 5.5% 2.2% 3.8% 1.1% 100%
4 Connected Country F1 Early-Adopting Elite Upscale Middle Age w/ Kids High-Tech Non-Target 13.9% 13.1% 3.9% 7.6% 8.6% 27.5% 6.8% 1.3% 13.8% 3.5% 100%
5 You & I Tunes Y1 Young & Wireless Upper Mid Younger w/o Kids High-Tech Magenta Primary 30.9% 27.5% 5.3% 5.0% 5.2% 9.2% 7.4% 1.1% 7.3% 1.1% 100%
6 High-Tech Society F1 Early-Adopting Elite Wealthy Older w/ Kids High-Tech Magenta Stretch 8.3% 18.8% 2.0% 2.2% 11.3% 19.3% 4.0% 0.3% 27.8% 5.9% 100%
7 Generation WiFi Y1 Young & Wireless Midscale Younger w/o Kids High-Tech Magenta Primary 33.9% 27.0% 6.4% 6.9% 5.8% 8.4% 5.3% 0.9% 4.3% 1.1% 100%
8 Calling Circles F2 Suburban Spenders Midscale Younger w/ Kids High-Tech Secondary Magenta 32.9% 13.4% 8.3% 16.2% 4.1% 12.1% 6.9% 2.3% 2.6% 1.2% 100%
9 Dish Country F2 Suburban Spenders Midscale Younger w/ Kids High-Tech Non-Target 29.2% 8.6% 6.6% 24.5% 4.1% 14.1% 7.3% 1.7% 2.9% 1.1% 100%
10 Smart Gamers F2 Suburban Spenders Upscale Older w/ Kids High-Tech Magenta Stretch 11.1% 14.2% 3.0% 3.8% 13.1% 16.7% 5.2% 1.2% 25.0% 6.7% 100%
11 WiFi Warriors M1 Wireless White-Collars Wealthy Middle Age w/o Kids High-Tech Magenta Primary 13.0% 15.2% 5.3% 5.3% 15.5% 17.7% 7.0% 0.8% 18.7% 1.6% 100%
12 Satellites & Silos Y1 Young & Wireless Upper Mid Middle Age w/o Kids High-Tech Non-Target 21.5% 10.7% 7.9% 9.5% 9.5% 10.4% 10.0% 1.8% 17.2% 1.5% 100%
13 Cyber Sophisticates M1 Wireless White-Collars Wealthy Middle Age w/o Kids High-Tech Magenta Stretch 9.2% 15.5% 4.1% 3.0% 15.0% 18.1% 6.7% 0.7% 25.4% 2.3% 100%
14 The Pragmatics Y1 Young & Wireless Upscale Middle Age w/o Kids High-Tech Magenta Primary 17.7% 15.8% 7.6% 8.7% 13.1% 15.6% 7.7% 1.2% 11.0% 1.5% 100%
15 Bundled Burbs F2 Suburban Spenders Upscale Middle Age w/ Kids High-Tech Magenta Primary 12.6% 12.1% 4.5% 8.8% 11.8% 31.0% 5.1% 1.1% 9.9% 3.1% 100%
16 Kids & Keyboards F2 Suburban Spenders Upper Mid Middle Age w/ Kids High-Tech Magenta Primary 17.7% 10.0% 7.1% 20.2% 6.4% 25.1% 5.9% 2.0% 4.0% 1.6% 100%
17 Time Shifters Y1 Young & Wireless Upper Mid Younger w/o Kids High-Tech Magenta Primary 23.2% 14.3% 9.9% 12.4% 8.8% 12.9% 8.9% 1.2% 7.7% 0.9% 100%
18 New Technorati Y2 Emerging Techies Upper Mid Older w/o Kids Mid-Tech Magenta Primary 24.2% 13.4% 11.9% 12.4% 7.6% 10.5% 10.5% 1.8% 6.9% 0.9% 100%
19 Gadgets Galore F3 Frugal Folks Upper Mid Older w/ Kids Mid-Tech Magenta Stretch 13.3% 8.2% 5.6% 7.6% 6.1% 9.7% 17.0% 3.6% 23.9% 5.1% 100%
20 Cinemaniacs M2 Mature Mid-Techs Upscale Older w/o Kids Mid-Tech Magenta Primary 15.4% 10.1% 9.9% 6.9% 12.1% 8.8% 13.7% 1.4% 19.8% 2.0% 100%
21 Multimedia Families F3 Frugal Folks Upper Mid Older w/ Kids Mid-Tech Secondary Magenta 21.3% 7.6% 11.7% 19.5% 3.8% 9.8% 14.4% 5.5% 4.6% 1.7% 100%
22 Analoggers M2 Mature Mid-Techs Wealthy Older w/o Kids Mid-Tech Non-Target 5.9% 8.6% 2.3% 2.1% 5.7% 7.6% 8.5% 0.3% 57.0% 2.0% 100%
23 Cyber Strivers F3 Frugal Folks Lower Mid Middle Age w/ Kids Mid-Tech Secondary Magenta 35.2% 7.6% 11.5% 22.5% 2.2% 5.9% 9.0% 3.9% 1.8% 0.4% 100%
24 Internet Hinterlands M2 Mature Mid-Techs Upscale Older Mostly w/o Kids Mid-Tech Non-Target 9.4% 4.8% 5.1% 4.9% 5.1% 4.3% 22.4% 1.9% 39.9% 2.4% 100%
25 Low-Speed Boomers M2 Mature Mid-Techs Wealthy Older w/o Kids Mid-Tech Non-Target 7.4% 7.7% 3.2% 2.5% 5.9% 6.8% 13.5% 0.7% 49.9% 2.4% 100%
26 Rural Transmissions M2 Mature Mid-Techs Midscale Older Mostly w/o Kids Mid-Tech Non-Target 18.8% 4.4% 14.1% 11.4% 5.0% 5.7% 25.8% 4.2% 9.0% 1.6% 100%
27 Video Vistas F3 Frugal Folks Lower Mid Middle Age w/ Kids Mid-Tech Secondary Magenta 30.2% 7.4% 11.6% 23.6% 2.6% 8.7% 9.5% 3.6% 1.9% 1.0% 100%
28 Big City, Small Tech F3 Frugal Folks Low Income Younger w/ Kids Mid-Tech Secondary Magenta 41.2% 7.6% 9.7% 22.6% 1.3% 3.9% 8.6% 3.6% 1.0% 0.4% 100%
29 IM Nation Y2 Emerging Techies Lower Mid Younger w/o Kids Mid-Tech Secondary Magenta 41.6% 11.4% 9.5% 12.5% 2.4% 6.0% 10.9% 2.1% 2.8% 0.7% 100%
30 Techs and the City Y2 Emerging Techies Lower Mid Younger w/o Kids Mid-Tech Secondary Magenta 38.7% 14.2% 9.7% 11.6% 3.7% 6.1% 10.9% 2.0% 2.9% 0.3% 100%
31 Plug & Play Y2 Emerging Techies Upper Mid Older w/o Kids Mid-Tech Secondary Magenta 23.9% 11.4% 14.4% 13.0% 8.2% 9.7% 11.3% 1.9% 5.5% 0.8% 100%
32 Family Dishes F3 Frugal Folks Lower Mid Middle Age Family Mix Mid-Tech Non-Target 29.6% 5.3% 14.3% 22.8% 1.9% 4.8% 14.3% 3.9% 2.6% 0.6% 100%
33 Digital Dreamers Y2 Emerging Techies Lower Mid Older w/o Kids Mid-Tech Secondary Magenta 29.9% 8.4% 14.0% 17.1% 3.7% 7.2% 12.7% 2.7% 3.6% 0.8% 100%
34 Gearing Up Y2 Emerging Techies Low Income Middle Age w/o Kids Mid-Tech Secondary Magenta 45.8% 11.2% 11.9% 11.8% 1.6% 4.2% 9.6% 1.7% 2.0% 0.2% 100%
35 Broadband Boulevards M2 Mature Mid-Techs Upper Mid Older w/o Kids Mid-Tech Magenta Stretch 11.2% 9.0% 6.0% 4.9% 5.4% 6.8% 22.1% 1.9% 30.4% 2.2% 100%
36 Opting Out M2 Mature Mid-Techs Upper Mid Older w/o Kids Mid-Tech Secondary Magenta 17.8% 6.5% 12.6% 7.4% 6.0% 6.3% 23.0% 2.5% 16.3% 1.6% 100%
37 Techtown Lites Y3 Young Startups Lower Mid Older Mostly w/o Kids Low-Tech Secondary Magenta 28.3% 6.0% 17.0% 17.0% 2.8% 3.6% 16.9% 4.3% 3.0% 1.2% 100%
38 New Kids on the Grid Y3 Young Startups Downscale Older Mostly w/o Kids Low-Tech Secondary Magenta 31.7% 9.3% 15.1% 12.8% 3.6% 4.9% 16.1% 3.0% 3.0% 0.4% 100%
39 Video Homebodies F3 Frugal Folks Low Income Older Mostly w/o Kids Mid-Tech Secondary Magenta 33.9% 6.7% 14.9% 19.5% 1.8% 3.7% 12.4% 4.3% 1.8% 0.9% 100%
40 Low-Tech Country Y3 Young Startups Downscale Older Mostly w/o Kids Low-Tech Non-Target 23.9% 3.8% 16.5% 12.2% 2.0% 2.6% 29.0% 6.7% 2.9% 0.5% 100%
41 Antenna Land M3 Offline Seniors Midscale Older w/o Kids Low-Tech Secondary Magenta 18.4% 5.7% 9.7% 9.0% 3.6% 5.2% 34.4% 3.6% 9.1% 1.4% 100%
42 Tech Skeptics M3 Offline Seniors Upper Mid Mature w/o Kids Low-Tech Non-Target 7.8% 5.4% 4.6% 3.9% 3.9% 5.2% 31.2% 1.8% 34.6% 1.7% 100%
43 Bucolic Basics M3 Offline Seniors Midscale Mature w/o Kids Low-Tech Non-Target 11.1% 2.9% 7.2% 6.2% 3.3% 3.2% 44.2% 2.8% 17.4% 1.7% 100%
44 Leisurely Adopters M3 Offline Seniors Low Income Older w/o Kids Low-Tech Secondary Magenta 26.6% 7.2% 13.6% 12.8% 2.0% 4.8% 26.1% 4.2% 2.3% 0.5% 100%
45 Landline Living M3 Offline Seniors Downscale Older Mostly w/o Kids Low-Tech Non-Target 24.9% 3.8% 12.4% 12.9% 1.5% 3.8% 32.9% 4.7% 2.8% 0.5% 100%
46 Old-Time Media M3 Offline Seniors Upper Mid Mature w/o Kids Low-Tech Non-Target 8.2% 6.0% 5.9% 4.0% 5.2% 6.8% 28.1% 1.7% 32.5% 1.6% 100%
47 Discounts & Deals M3 Offline Seniors Downscale Older Mostly w/o Kids Low-Tech Non-Target 22.5% 6.1% 10.6% 10.3% 2.5% 4.2% 30.9% 4.4% 7.4% 1.1% 100%
48 Dial-Up Duos M3 Offline Seniors Midscale Mature Mostly w/o Kids Low-Tech Non-Target 11.6% 4.4% 7.8% 6.9% 3.3% 4.1% 44.8% 3.2% 12.6% 1.3% 100%
49 Satellite Seniors M3 Offline Seniors Downscale Older Mostly w/o Kids Low-Tech Non-Target 19.0% 2.6% 9.3% 12.4% 1.1% 3.7% 42.4% 4.7% 4.4% 0.3% 100%
50 Early-Bird TV M4 Elderly Traditionalists Midscale Mature w/o Kids No-Tech Non-Target 10.6% 5.7% 7.5% 5.7% 4.1% 5.1% 42.2% 1.3% 16.5% 1.4% 100%
51 Tech-Free Frontier M4 Elderly Traditionalists Downscale Mature w/o Kids No-Tech Non-Target 16.7% 3.6% 10.5% 9.9% 2.1% 3.0% 43.6% 2.6% 7.6% 0.4% 100%
52 The Unconnected M4 Elderly Traditionalists Downscale Mature w/o Kids No-Tech Non-Target 17.0% 5.9% 10.4% 9.9% 3.4% 5.4% 36.4% 3.2% 7.5% 0.9% 100%
53 Last to Adopt M4 Elderly Traditionalists Low Income Older w/o Kids No-Tech Non-Target 23.9% 6.1% 11.8% 11.9% 2.1% 4.8% 33.1% 3.4% 2.6% 0.4% 100%
99 Unclassified 99 Unassigned Unassigned Unassigned Non-Target 36.1% 18.3% 6.6% 13.3% 3.0% 10.6% 6.2% 0.8% 4.0% 1.1% 100%
Source: In-house T-Mobile Depiction
D2C| SEGMENTATION,
TARGETING
Usage Example: Segment Demographic and Lifestyle Profiling (Segment 2)
8.2% 12.3%
US HHs US GAs
50% 50%
UNDERSTAND ENGAGE
AGE OF HEAD OF HOUSEHOLD HH INCOME LIFESTYLE
5% <$75k
67% 14% Lead an active lifestyle – enjoy soccer, tennis, baseball, softball, basketball,
40% $75k-$100k football, jogging, golf, biking, and yoga
33%
$100k-$150k
0% 0% 0% 0% Prefer To Shop At Specialized Stores
$150k-$250k
40%
18-24 25-34 35-44 45-54 55-64 65+ $250k+ Rather Book a Trip over the Internet than Meet Travel Agent
41% Use smartphone for fantasy sports,

RACE & ETHNICITY travel reservations, live sports, traffic
EDUCATION 24%
Caucasian
60 13%
21%
Use tablet for fantasy sports, watching
TV, blogs, sports scores, movies
Hispanic
Asian
17%
11%
% MOBILE
DEVICE
African American 9%
Other/Decline 4% High School Some College USAGE Use social network to keep in touch with
Hispanic ethnicity is not mutually exclusive of race Undergraduate Postgraduate friends /family and for prof. networking
95% own smartphone and 48% own tablet;

42% handset brands: 51% Apple, 26% Samsung
# OF KIDS 24% 23% “I think of my mobile phone as a source of entertainment.”
AT HOME
7% 4% 70% ATTITUDES
“I am interested in watching TV/video clips on my cell phone.”
“My mobile device is an extension of my personality.”
Home Owners “I would feel disconnected without the Internet.”
0 1 2 3 4+ “I like to live a lifestyle that impresses others.”
“Most trusted media is Internet. I would feel disconnected without the Internet”
Source: Nielsen Mobile Insights Source: Nielsen ConneXions 2016; Nielsen Demographics 2016; Nielsen Consumer Profiles 2016; GfK MRI 2016
D2C| SEGMENTATION,
TARGETING
Usage Example: Gross Adds Tracking Click icon to open
report Segment GAs
 Overall Gross Adds (those who’ve switched somewhere within the past year) show shifts over time in the
Millennial segments
Total GA Segment Movement If referred to simply as “gross adds” you can

assume they switched in the past year at
40.0%
some point
35.0%
30.0%
25.0%
20.0%
15.0%
10.0%
5.0%
0.0%
Q1 2013 Q2 2013 Q3 2013 Q4 2013 Q1 2014 Q2 2014 Q3 2014 Q4 2014 Q1 2015 Q2 2015 Q3 2015 Q4 2015 Q1 2016 Q2 2016 Q3 2016 Q4 2016 Q1 2017 Q2 2017
Source: Nielsen Mobile Insights, blended postpaid and prepaid
D2C| SEGMENTATION,
TARGETING
Usage Example: Segmentation and Promos (T-Mobile Subscribers)
 T-Mobile participation is dominated by $0 down, followed by device trade-in and device discounts
T-Mobile Segments - Recent Promo Participation

Base: T-Mobile Subscribers Responsible for Device Selection Within Past 6 Months
$0 down for new device Instant POS rebate Service bundling Device discounts Free Device BOGO Mail-in rebate Bundled accessories
Percent of Promo Participants
Device trade-in special Switching credit Online specials Value added promos Gift card offering Additional airtime Other promotion
25%
20%
15%
10%
5%
0%
1 2 3 4 5 6 7 8 9 10
Segments
Source: Nielsen Mobile Insights, Q4 2015 to Q4 2016
D2C| SEGMENTATION,
TARGETING
Usage Example: Segmentation and Promos (Non-T-Mobile Subscribers)
 Non-T-Mobile participation is led by device discounts and device trade-in, followed by $0 down
Non-T-Mobile Segments - Recent Promo Participation

Base: Non-T-Mobile Subscribers Responsible for Device Selection Within Past 6 Months
$0 down for new device Instant POS rebate Service bundling Device discounts Free Device BOGO Mail-in rebate Bundled accessories
Percent of Promo Participants
Device trade-in special Switching credit Online specials Value added promos Gift card offering Additional airtime Other promotion
25%
20%
15%
10%
5%
0%
1 2 3 4 5 6 7 8 9 10
Segments
D2C| SEGMENTATION,
TARGETING
Usage Example: Products and Feature Adoption (TMO vs. Market)
 TMO significantly outperforms market on the customer adoption of the following features: Unlimited data, Wi-Fi calling, Free video and music streaming and no contra
 TMO also leads market roaming (more so in North America) but the competitive lead is lower
Product Features by Segment: TMO vs. Market

Device
Free calling to
Unlimited high- Wi-Fi Free global Free streaming of Free tethering / No service Device equipment
Canada and
speed LTE data calling roaming video and music Wi-Fi hotspot contract financing protection/
Mexico
upgrade plans
TMO Non-TMO TMO Non-TMO TMO Non-TMO TMO Non-TMO TMO Non-TMO TMO Non-TMO TMO Non-TMO TMO Non-TMO TMO Non-TMO
Segment 1: Millennials < $75K 72% 44% 23% 18% 66% 45% 15% 12% 50% 27% 52% 29% 46% 39% 43% 32% 40% 38%
Segment 2: Millennials > $75K 58% 39% 27% 13% 61% 46% 23% 10% 46% 21% 32% 27% 37% 25% 32% 41% 26% 45%
Segment 3: GenX < $75K no kids 87% 43% 19% 21% 74% 40% 20% 12% 50% 25% 42% 29% 43% 48% 48% 23% 32% 30%
Segment 4: GenX < $75K with kids 74% 45% 22% 19% 70% 37% 15% 10% 51% 27% 33% 33% 52% 47% 52% 33% 45% 37%
Segment 5: GenX > $75K no kids 70% 40% 25% 17% 65% 31% 20% 11% 46% 19% 40% 23% 53% 24% 48% 40% 36% 43%
Segment 6: GenX > $75K with kids 79% 42% 39% 19% 78% 36% 21% 9% 64% 21% 45% 29% 48% 23% 54% 45% 49% 37%
Segment 7: Boomers+ < $75K no kids 61% 34% 26% 25% 58% 31% 14% 16% 35% 16% 36% 21% 65% 46% 36% 25% 30% 24%
Segment 8: Boomers+ < $75K w/ kids 82% 36% 49% 16% 79% 39% 9% 10% 49% 17% 48% 25% 78% 40% 40% 29% 24% 35%
Segment 9: Boomers+ > $75K no kids 70% 30% 33% 23% 59% 35% 12% 8% 45% 16% 38% 25% 58% 25% 34% 34% 23% 29%
Segment 10: Boomers+ > $75K w/ kids 63% 33% 35% 17% 67% 38% 13% 12% 40% 26% 46% 28% 73% 24% 70% 43% 36% 42%
Total 71% 39% 26% 20% 66% 38% 17% 12% 48% 21% 41% 27% 51% 37% 43% 33% 36% 34%
Index of TMO to Market by Segment

Device
Free
Unlimited Free calling Free equipment
Wi-Fi Free global streaming of No service Device
high-speed to Canada tethering / Wi- protection/
calling roaming video and contract financing
LTE data and Mexico Fi hotspot upgrade
music
plans
Segment 1: Millennials < $75K 128 105 122 103 123 123 107 112 102
Segment 2: Millennials > $75K 119 114 115 114 125 106 111 91 81
Segment 3: GenX < $75K no kids 144 99 134 108 125 113 94 125 102
Segment 4: GenX < $75K with kids 129 103 134 105 123 100 105 120 108
Segment 5: GenX > $75K no kids 131 108 134 109 127 117 129 108 93
Segment 6: GenX > $75K with kids 136 120 143 112 143 116 124 109 112
Segment 7: Boomers+ < $75K no kids 127 101 126 98 119 114 118 111 106
Segment 8: Boomers+ < $75K w/ kids 147 133 140 99 132 122 137 111 89
Segment 9: Boomers+ > $75K no kids 140 110 124 104 129 112 132 100 93
Segment 10: Boomers+ > $75K w/ kids 130 118 130 100 114 117 148 127 94
Total 132 106 129 105 127 115 114 111 102
D2C| SEGMENTATION,
TARGETING
Usage Example: LA Local Market Analysis
Note: Segments 7 to 10 (Boomers) are suppressed in this map
D2C| SEGMENTATION,
TARGETING
Usage Example: NYC Local Market Analysis
Note: Segments 7 to 10 (Boomers) are suppressed in this map
D2C| SEGMENTATION,
TARGETING
Goatfell Market Opportunity Sizing
Total US Pop
325,040,390
Boomers 55+
140,751,033
Boomers 55+ Prospects

129,894,771
Segments 7/9 Segments 8/10

70,136,457 60,033,006
Segments 8/10 <= 2

Addressable Opportunity Segments 8/10 > 2 Lines
Lines
106,127,576 24,041,887
35,991,118
Mid-High Incrementality
Low Incrementality
(TMO Mkt Share < 12% 7/9 or < 15% 8/10)
(TMO Mkt Share > 12% 7/9 or > 15% 8/10)
9,568,683
96,558,892
Group A: Group B: Group C: Group D:

Strength Markets Opportunity Markets Emerging Markets Greenfield Markets
21,249,607 54,524,368 9,410,303 11,374,614
Urban Suburban S’Town/Rural Urban Suburban S’Town/Rural Urban Suburban S’Town/Rural Urban Suburban S’Town/Rural
10,506,599 10,718,262 24,746 6,403,942 37,054,333 11,066,093 247,117 4,642,832 4,520,354 13,516 4,916,965 6,444,133
D2C| SEGMENTATION,
TARGETING
Limitations of the Current Demographic Segmentation
 Uses simple rule-defined syntax instead of a predictive or pattern-discerning algorithm.
 Excludes key predictive variables such as wireless, lifestyle and media consumption behaviors, and
as such has limited predictive power.
 Based on presumption that demographics alone (age, income and presence of kids) are the key
differentiators of consumer wireless behavior; it thus ignores other salient differentiating variables.
 Assumes that members of each demographic cohort have similar preferences and same behavioral
orientation, ignoring the inherent differences within these cohorts.
 Difficult to match to online behaviors (since demographics are not reliable delineators of web
behavior). As such, it is of limited use for digital targeting.
 Whilst salient within T-Mobile’s customer base, it is not necessarily a “wireless” segmentation that
can be used in a non-wireless prospect file for campaign selection.
D2C| SEGMENTATION,
TARGETING
Where We Are Going: Conceptual Approaches
D2C| SEGMENTATION,
TARGETING
Summary of New Segmentation Objectives and Options
Modeling Use as Predictive Model Summary of Pros & Cons
Algorithm-based Offer Channel Message Media
Objectives Analysis Plan Mkt. Cov. Prediction Resource
Behavioral Model • Market segmentation • Use core inputs
Leverages advanced algorithm • Target development • Other multi-sourced
and wide range of •
demographic, behavioral and
psychographic variables to
•
Product differentiation
Positioning/Messaging     •
inputs
Use random tree ● ◕ ◐
build market-level • Sizing/segment algorithm
segmentation economics • Determine target variable
• Append base CLVs
Customer CLV Model • Target development • Predict prospect CLVs
Extends TMO CLV model to • Product differentiation • Create new CLV cohorts
general market, using this as • Positioning/Messaging
    • Use cohort as model
◐ ◔ ◔
Power
Predictive Power
target variable for segmentation

• Sizing/segment target
based on lifetime spend and the
variables driving such spend economics • Multinomial logistic
model
Brand Affinity Model • Market classification • Pre-test input validity
Discriminant model using • Target development • Append carrier brand
curated characteristics of TMO
◔ ◔ ◔
• •
   
Product differentiation Use cohort as model
base to identify potential targets
in the broader market versus
• Positioning/Messaging target
likely non-customer types • Sizing/segment • CHAID decision tree
economics model
Rule-based
New Demographic Model • Append additional demos
◔ ◔ ◔
•
Appends additional variables to
DTC’s existing demographic
segments, but retains the same
•
Target development
Messaging     •
•
Use rule-based logic
No algorithm
rule-based logic
D2C| SEGMENTATION,
TARGETING
Analysis Plan Details
Modeling Sample Input Data Model Specifications
Source Size Coverage Category Level Source Availability Readiness Description Output Deployment
Prospect Random Forest Once the
Individual/HH Merkle  ◕ Among
demographics model is initial model
other
Prospect Merkle or specified, as a is build, a
Individual/HH  ◕ outputs,
• Total sample: 1m lifestyles Claritas supervised scoring
random
individuals Training sample will Prospect life segmentation syntax based
forests
• Prospect cover the entire US events
Individual/HH Merkle   technique based
produce
on
base • 80% prospect wireless market, drawn on recursive commonly
Behavioral Credit card group
Model
in random proportions
purchases
Individual Merkle   partitioning.
membershi
available
• Custome • 20% TMO, of to reflect TMO versus variables will
Customer data Individual EDW  ◕ p and
r base which postpaid = competitor market Random forest be built to
probability
90% and 10% = share Customer web Engineerin algorithm will enable the
Individual  ◕ scores
prepaid behavior g support optimal deployment
enabling
data reduction of the model
Algorithm-based
segment
Customer MET CLV and multivariate to population
Individual
Model
 ◔ segmentation
allocation.
platforms
lifetime value
Customer Multinomial Among
Individual/HH Merkle  ◕ Scoring
Training sample will be demographics logistic other
• Total sample: Customer data Individual EDW  ◕ syntax using
drawn from TMO regression is results,
500k common
Customer • Custome customer base, with specified, since model
Engineerin variables will
CLV Model r base post-modeling scoring Customer web Individual  ◕ we have a target generates
• 90% postpaid, behavior g be built to
applied to prospect variable – CLV grouping &
and 10% prepaid append
base Customer MET CLV cluster probability
Individual  ◔ segmentation
lifetime value Model membership scores
Customer Model
Training sample will be demographics Individual/HH Merkle  ◕ CHAID decision
• Total sample: appends
Brand drawn from TMO tree model for Scoring
• Custome 500k Customer data Individual EDW  ◕ node ID,
customer base, with input syntax used
Affinity r base
post-modeling scoring partitioning,
info gain
to append
Model • 90% postpaid, Customer web Engineerin table &
applied to prospect Individual  ◕ with TMO base segmentation
and 10% prepaid behavior g prob.
base as target variable
scores
Training sample will be Customer A rule-based
• Total sample: Individual/HH Merkle  ◕ Model Same rule-
Rule-based
drawn from TMO demographics model is

New • Custome 500k appends based syntax
customer base, with Customer data Individual EDW  ◕ specified based
Demographi r base
post-modeling scoring on
grouping is used to
c Model • 90% postpaid, Customer web Engineerin code based append
applied to prospect Individual  ◕ predetermined
and 10% prepaid behavior g on rules segmentation
base parameters
D2C| SEGMENTATION,
TARGETING
Model Selection
 Model selection criteria:
— Actionability and alignment with current marketing strategy
— Relevance of the modeling sample and input variables to the general wireless market
— Predictive value to support offer, channel, messaging and media buying decisions
— Amount of resources needed to develop and implement model
 Proposed approach:
— Combine diagnostic test with the behavioral and brand affinity models
 Justification
— A combination of variable pre-testing with targeted behavioral and brand affinity models allows for a holistic
data input, effective data reduction, and highly optimized segmentation
— Finalizing with a brand affinity model helps to narrow segmentation inputs to an actionable range
D2C| SEGMENTATION,
TARGETING
Diagnostic Test
 Analysis Objective:
— Objective: Determine the relationships among input variables and assess the predictive power of input data against named
targets (targets = TMO Membership in MDB and Likelihood to Switch in MI)
 Testing Methods:
— Data Web: A data depiction method to quantify the frequency of co-occurrence between specified variables, using network
graph and the output data.
— Correlation/PCA: A correlation matrix to quantify the relationship between variables, with orthogonal transformation to
convert correlated variables into a set of values of uncorrelated variables called principal components or factors. The
components could be used as discrete inputs in our segmentation modeling.
— ROC Curve: The ROC curve is a fundamental tool for diagnostic test evaluation in which the true positive rate (Sensitivity) is
plotted in function of the false positive rate (100-Specificity) for different cut-off points of a parameter. Each point on the
ROC curve represents a sensitivity/specificity pair corresponding to a particular decision threshold. The area under the ROC
curve (AUC) is a measure of how well a parameter can distinguish between two diagnostic groups, in our case TMO
membership or Likely Switchers.
 Expected Results
— Diagnostic testing will enable us to reduce input variables by (a) assessing their relationships and using this to create
component variables; and (b) by establishing input cutoffs through estimates of the predictive power of input variables
D2C| SEGMENTATION,
TARGETING
Data Web Example
D2C| SEGMENTATION,
TARGETING
Correlations/Principal Components Example
Inte r-Ite m Co rre latio n Matrix
Voice network Data network Monthly Overall

Overall quality quality spend on Cust. Service Satisfaction
Presence Likely Credit Carrier Satisfaction satisfaction satisfaction cell phone Satisfaction with Handset Geo-
Age of kids HH Income switcher Proxy Plantype Tenure (Top 3 Box %) (Top 3 Box %) (Top 3 Box %) bill (Top 3 Box %) (Top 3 Box %) density RaceEthn
Age 1.000 -0.359 0.142 -0.161 0.221 0.093 0.271 -0.011 -0.064 -0.072 -0.041 -0.029 0.015 -0.026 -0.030
Presence of kids -0.359 1.000 0.075 0.153 -0.124 -0.050 -0.166 -0.019 0.000 -0.001 0.113 -0.020 -0.035 0.020 -0.006
HH Income 0.142 0.075 1.000 0.058 0.298 0.183 0.132 -0.011 -0.049 -0.073 0.200 -0.019 -0.085 -0.009 -0.037
Likely switcher -0.161 0.153 0.058 1.000 -0.062 -0.194 -0.284 0.185 0.126 0.124 -0.013 0.116 0.042 0.011 -0.018
Credit Proxy 0.221 -0.124 0.298 -0.062 1.000 0.167 0.150 0.001 -0.023 -0.032 0.038 -0.001 -0.032 -0.018 -0.019
Plantype 0.093 -0.050 0.183 -0.194 0.167 1.000 0.370 0.081 0.002 0.000 0.428 0.034 -0.071 -0.014 0.007
Carrier Tenure 0.271 -0.166 0.132 -0.284 0.150 0.370 1.000 0.002 -0.084 -0.067 0.205 -0.006 -0.054 -0.020 0.007
Overall Satisfaction (Top 3 Box %) -0.011 -0.019 -0.011 0.185 0.001 0.081 0.002 1.000 0.608 0.626 0.077 0.645 0.370 -0.003 0.029
Voice network quality satisfaction (Top 3 Box %) -0.064 0.000 -0.049 0.126 -0.023 0.002 -0.084 0.608 1.000 0.673 0.005 0.495 0.356 0.000 0.026
Data network quality satisfaction (Top 3 Box %) -0.072 -0.001 -0.073 0.124 -0.032 0.000 -0.067 0.626 0.673 1.000 -0.002 0.499 0.363 0.003 0.035
Monthly spend on cell phone bill -0.041 0.113 0.200 -0.013 0.038 0.428 0.205 0.077 0.005 -0.002 1.000 0.039 -0.071 0.004 -0.006
Cust. Service Satisfaction (Top 3 Box %) -0.029 -0.020 -0.019 0.116 -0.001 0.034 -0.006 0.645 0.495 0.499 0.039 1.000 0.361 -0.004 0.024
Overall Satisfaction with Handset (Top 3 Box %) 0.015 -0.035 -0.085 0.042 -0.032 -0.071 -0.054 0.370 0.356 0.363 -0.071 0.361 1.000 -0.002 0.025
Geo-density -0.026 0.020 -0.009 0.011 -0.018 -0.014 -0.020 -0.003 0.000 0.003 0.004 -0.004 -0.002 1.000 0.008
RaceEthn -0.030 -0.006 -0.037 -0.018 -0.019 0.007 0.007 0.029 0.026 0.035 -0.006 0.024 0.025 0.008 1.000
Component Matrix
1 2 3 4 5
Presence of kids 0.703 -0.283 0.266 -0.001 -0.004
Q210r 0.602 0.502 -0.190 0.005 -0.047
Age -0.614 0.486 0.015 -0.095 0.028
Plantype 0.335 0.692 -0.277 -0.010 -0.042
Carrier Tenure -0.070 0.659 -0.346 -0.103 -0.035
HH Income 0.197 0.498 0.597 -0.109 0.104
Credit Proxy -0.197 0.510 0.529 -0.066 0.135
Overall Satisfaction 0.032 0.155 0.006 0.845 -0.005
Voice Quality Satisfaction 0.014 0.017 0.053 0.821 0.001
Data Quality Satisfaction 0.009 0.011 0.018 0.830 0.003
Cust. Service Satisfaction 0.004 0.111 0.021 0.771 -0.007
Overall Handset Satisfaction -0.182 -0.054 0.010 0.585 0.008
RaceEthn 0.014 -0.027 -0.383 0.056 0.347
Geo-density 0.068 -0.052 -0.035 0.002 0.919
Extraction Method: Principal Component Analysis.
D2C| SEGMENTATION,
TARGETING
ROC Curve – TMO Membership
D2C| SEGMENTATION,
TARGETING
ROC Curve – Combined TMO and Metro Membership
D2C| SEGMENTATION,
TARGETING
ROC Curve – Likely to Switch
D2C| SEGMENTATION,
TARGETING
ROC Curve – Not Likely to Switch
D2C| SEGMENTATION,
TARGETING
Brand Affinity Model
— Identify and rank the variables driving consumer affinity for TMO, and use the key drivers to
increase TMO’s penetration of base-like market segments (viz: feature development and
market penetration strategy)
 Modeling Approach:
— CHAID algorithm: A statistical method used for detection of interaction between variables,
revealing how variables are best merged to predict discrete outcomes or values in a given
dependent variable. CHAID is a non-parametric decision tree technique based on adjusted
significance testing. It can handle nominal, ordinal and continuous data; and in multi-way partitioning creates all possible cross tabulations for
each categorical predictor until the best outcome is achieved and no further splitting can be performed. In the CHAID technique, we can
visually see the relationships between the split variables and the associated factor within the tree. The development of the decision, or
classification tree, starts with identifying the target variable or dependent variable, which
would be considered the “root”. CHAID analysis splits the target into two or more categories
that are called the “initial”, or “parent” nodes, and then the nodes are split using statistical
algorithms into “child” nodes. Unlike regression analysis, CHAID does not require data to be
normally distributed. It is therefore non-parametric. This is suitable for TMO’s CRM data.
— Our CHAID analysis will yield the following appended or reported output:
— Node ID: A membership variable appended to the dataset showing terminal node for each record
— TMO probability: Probability of predicted TMO membership based on input variables
— Non-TMO probability: Probability of non-TMO membership based on input variables
— Information gain table: A rank-ordered table measuring the utility of inputs in discriminating between the classes (i.e. reduce impurity) in the target variable
— Classification rules: A syntax showing rules of assignment to nodes – useful for developing a post-hoc typing tool
— Classification table: This is a table showing the binary target observations used to build the model comparable to the predicted binary outcomes. The bar
chart above compares our observed and predicted classification outcomes using Mobile Insights survey data as a test case.
D2C| SEGMENTATION,
TARGETING
Behavioral Segmentation
— Leverage wide range of input variables to build a market segmentation solution that covers
whole-market behaviors and motivations, and use it to enlarge total addressable market for
TMO (viz: market diversification strategy)
 Modeling Approach:
— Random Forest algorithm: This is a statistical method which uses an aggregation or
ensemble of decision trees to cope with large sets of input variables. Standard decision tree
algorithms are strained by the complexity of data interactions inherent in large input arrays,
resulting in overfitting (over-complex partitioning) and information loss. Random Forest uses a stochastic discrimination approach to correct
for weak learning by partitioning the array of inputs into random subsets, each subset optimized by a random predictor variable. In other
words, the eligible predictor set will be quite different from node to node. When a new input is entered into the system, it is run down all of
the trees. The result may either be an average or weighted average of all of the terminal
nodes that are reached, or, in the case of categorical variables, a voting majority. An optimal
Random Forest outcome should allow minimal correlation between subset trees, since the
greater the inter-tree correlation, the greater the random forest error rate. A Random Forest
approach will enable us to fit the disparate array of data from our CRM and survey sources.
— Our Random Forest analysis will yield the following appended or reported output:
— Node ID: A membership variable appended to the dataset showing terminal node for each record
— TMO probability: Probability of predicted TMO membership based on input variables
— Non-TMO probability: Probability of non-TMO membership based on input variables
— Information gain table: A rank-ordered table measuring the utility of inputs in discriminating between the classes (i.e. reduce impurity) in the target variable
— Classification rules: A syntax showing rules for assignment to nodes – useful for developing a post-hoc typing tool
— Classification table: This is a table showing the binary target observations used to build the model comparable to the predicted binary outcomes. The bar
chart above compares our observed and predicted classification outcomes using Mobile Insights survey data as a test case.
D2C| SEGMENTATION,
TARGETING
Where We Are Going: Implementation
D2C| SEGMENTATION,
TARGETING
Development and Activation of Behavioral Segmentation
Inputs* Method Deliverables
• T-Mobile Base Behaviors • Normalize all datasets into single in-house • Scored Base and prospect file and
data lab subsequent onboarding to digital ecosystem
• T-Mobile Web & Channel
Behaviors • Implement Cluster/Tree/Neural Network • Magenta People
method and create numerous scenarios
• Wireless Behaviors and • Distributions (Geographic counts) at BG,
with various # of segment outcomes (ex: 5,
Motivations Zip and DMA levels
10, 15)
• Scarborough Lifestyle and Media • Typing tool for Primary Research and
• Review select few top performing models
usage Behaviors anonymous website visitors where
with cross functional team to select best fit
cookies/Device Ids can’t be matched
• Demographics • Development of algorithm/process and
typing tool to implement new segments
onto marketing file and in web ecosystem
D2C| SEGMENTATION,
TARGETING
Behavioral Segmentation: Summary of Required Data Inputs
Prospect T-Mobile Base T-Mobile Web Mobile Insights Media Consumption
Demos/Behaviors Behaviors Behaviors Variables
• Person demographics • Current plan type: • Mobile/non-mobile website • Current carrier • Television
‒ Age category (18+ only) ‒ Individual postpaid visits • Current plan type: ‒ Viewing by genre
‒ Gender ‒ Family plan postpaid • Multi-page visits ‒ Individual postpaid ‒ Viewing past 7 days
‒ Race** ‒ Pre-paid • Page depth ‒ Family plan postpaid o Broadcast
‒ Ethnicity** • Plan type (e.g.)*: ‒ Prepaid: by minute o Cable networks
• Average clicks per visit o Specialty
‒ Position in HH ‒ # of lines per plan ‒ Prepaid: by month with
‒ Education level** • Average page views per visit programming
‒ Y/N kickback & value back unlimited minutes
‒ Occupation** ‒ T-Mobile One • Average time spent per page ‒ Prepaid: annual fee (awards)
• Person’s Household Characteristics • Average time spent per visit o Specialty sports
‒ T-Mobile One Plus
• Average clicks per page • Plan and device usage:
‒ Income ‒ T-Mobile One International ‒ Handset purchaser ‘loyal programming
‒ Presence/Age of Children ‒ # of Tablets on plan • Number of repeat visits upgrader’ ‒ Devices and attributes
‒ Urbanicity ‒ # of wearables on plan • Average time between visits ‒ Handset purchaser ‘recent o Cable type
‒ Other VARs to be considered: ‒ No credit check • Non-shopping page visits switcher’ o DVR
o Credit status ‒ Pay as you go plan types • Shopping page visits ‒ Handset purchaser: ‘New to o Pay per
o Net Worth • Offer accepted at sign up • User Journey wireless’ view/VOD
o Own/Rent (e.g.)*: ‒ Entry page ‒ High-end / ultra high-end • Internet
• Lifestyles ‒ Third line free ‒ Engaged page phone owner ‒ Top news, sports and
• Life events ‒ iPhone 7 free ‒ Exit page ‒ Likely to upgrade in 6 entertainment sites
• Purchase behaviors ‒ Free Samsung Galaxy • Campaign impressions months or less • Newspapers
‒ DUP/IEP ‒ Daily reader - print
• ConneXions • Individual device usage (e.g.)*: • Campaign click-throughs
edition – national papers
‒ % of 4G network • Ad impressions ‒ 4g network
‒ % using data apps ‒ Unlimited voice/text ‒ Newspaper sections read
• Click to chat ‒ Mobile device used to
‒ Video usage where ‒ Using data apps
• Info page read
possible • Cart starts • Behaviors
‒ Etc. ‒ Likely switcher, reason • Radio
• Cart removals considering switch ‒ By format
• CLV • Web order submit ‒ By station ownership
‒ Lifetime spend ‒ Considering T-Mobile
‒ Non-switcher reasons
‒ Reason left previous carrier
• Channels (web, phone, mobile)
• Psychographics
*All behaviors will be reviewed and jointly agreed by key stakeholders prior to modeling process **Usability will depend on fill rate available for all persons; may use HH variable where person is not present on file
D2C| SEGMENTATION,
TARGETING
Data Inputs: Sources, Status and Prioritization
Data Category Level Source Priority Availability Comment
In-house or Readiness for
Contracted Sample Use
Individual/H
Prospect demographics H Merkle ●  ◕ Part of core prospect sample
Prospect File
Individual/H ◕
Prospect lifestyles H
Merkle or Claritas ●  Claritas requires append
Individual/H
Prospect life events H Merkle ●   Expected Merkle module
Credit card purchases Individual Merkle ●   Expected Merkle module

Customer data Individual EDW ●  ◕ Part of core base sample
Customer File
Customer web behavior Individual Engineering ●  ◕ Dependency on Engineering

Dependency on Portfolio
Customer lifetime value Individual MET CLV Model ●  ◔ team
Wireless behaviors (Mobile Individual/H Nielsen/Stagwell
Insights) H Group
●  ◔ Requires append
Already appended to
ConneXions Code Household Claritas ◕  ◐ prospect
Other Third Party
Individual/H
General purchase behavior H
Claritas ◐  ◔ Requires append
Prospect tech behavior Individual Claritas ◐  ◔ Requires append

Input Priority: Input transformation ◔Requires modeling Not available
Prospect ● High
media ◕Medium ◐ LowIndividual/H
consumption NielsenReadiness:
◕Available to◕pull ◐Needs
Scarborough   Needs to be purchased
H
D2C| SEGMENTATION,
TARGETING
Model Schematic: Data Sources, Development Plan and Deployment Targets
Conceptual diagram showing input and output channels for the behavioral segmentation
Data Analysis Engines

IDW Customer Data: Model development:
• Account/Subscriber ID SQL Assistant • Data prep
• Promotions/Features
• Initial model build
• Customer behavior data SAS, SPSS, R • Testing and optimization
• CDR
Tableau • Model validation
• Online behavior
Claritas ConneXions:
• Demographics
• Lifestyle
• Psychographics Jolt Data Lab:
• Wireless behavior • Customer data
• Online behavior • Prospect data Final
Final
• Media consumption • Telesales Mode
Mode
Merkle Prospect data: • Web behavior ll
• Demographics • Media
• Lifestyle • Campaign data
• Psychographics
• Wireless behavior
• Online behavior
Nielsen Mobile • ConneXions
Insights:
• Demographics AAM Platform: Adobe Workbench:
• Wireless purchase • Campaign analysis
• On-boarded Merkle data
experience • Predictive analytics
• Cookie data
• Device preference and usage • Visualization
• ConneXions
• Customer experience • Retail data
• Satisfaction drivers
• Switching and consideration
Channel Key Secondary

Existing inputs SiteCat: Deployments
• Cookie data PrimeLocation/E5
Expected inputs • Basic campaign reporting SMS/Scarborough
• Sundry metrics
Output channel Custom Surveys
Deployment channels
D2C| SEGMENTATION,
TARGETING
Brand Affinity Model: “Magenta People” Profile Development Process
 Once behavioral segmentation is completed, we will develop a Magenta People construct using core, actionable
inputs
Magenta Core (Media) Magenta Core (Non Media) Magenta Stretch

High-value segments showing
Mid- to high-value segments Low-value segments showing
strong Magenta affinity similar
showing high Magenta affinity high Magenta affinity similar to
to prime segments in current
similar to current base core current non-core base
base
Magenta People
Magenta People
Profiles
DEMOGRAPHICS LIFESTYLE ATTITUDE TECHNOLOGY MOBILE SOCIAL MEDIA ONLINE
 Affinity model inputs will be determined through testing and optimization

D2C| SEGMENTATION,
TARGETING
Behavioral Segmentation: Current Status and Dependencies
Phase Milestone Status Dependencies Suggested Timeframe
Step 1 Input Processing/Evaluation TBD Leadership approval September 2017
Step 2 Data Migration to Jolt TBD MET workload October 2017
Step 3 Segmentation Dataset Prep TBD Jolt readiness November 2017
Step 4 Behavioral Segmentation Build TBD Jolt readiness Jan-Mar 2018
Support of platform
Step 5 Magenta Affinity Model Devt TBD
owners and MET team
April/May 2018
Step 6 Testing and Deployment TBD Segment completion June-Sep 2018
D2C| SEGMENTATION,
TARGETING
Next Steps
 Project socialization
 Develop team-level usage scenarios
 Socialize segmentation plan
 Conduct technical consultations
 Sample data processing

 Sample development
 Data cleansing
 Data migration to Jolt
D2C| SEGMENTATION,
TARGETING
Appendix
D2C| SEGMENTATION,
TARGETING
Merkle DataSourceSM for Consumer
Wealth
Transactional indicators Real
Real property
property
Transactional Syndicated
Mortgage
Mortgage cooperative data
data Social media
cooperative data
data Direct
Direct integration
integration with
with
handles/flags / research data
social influence Facebook, Twitter,
Facebook, Twitter, AOL,
AOL,
Summarized Google,
Google, Rentrak,
Rentrak, etc
etc
credit statistics
Lifestyle
Lifestyle // behavioral
behavioral
Life events & triggers Segmentation
Segmentation Cookie
New
New movers
movers Systems (examples)
Systems (examples) synchronized
New
New homeowners
homeowners Mosaic
Mosaic Merkle for display to standard & Syndicated to Oracle/Bluekai,
New
New parents
parents
Automotive
Automotive data
data PRIZM Merkle mega,
mega, derived
derived Syndicated to Oracle/Bluekai,
PRIZM
& custom
Tri-bureau
Tri-bureau credit
credit models
models &
& summarized
summarized CAMEO
CAMEO & propensity
propensity RocketFuel,
RocketFuel, Google,
Google, Adobe,
Adobe,
Demographics variables audiences AOL
auto
auto statistics
statistics variables AOL
30+ 2500+ Daily new Direct linkage keys with Monthly new 95%+ 90-95%
major data selectable data movers Audience Platforms & homeowners coverage of all U.S. match rates on data
sources elements Datalogix/Oracle households overlay applications
129MM 275MM 500MM 475MM 130MM 291MM 75MM

households validated across individuals email phones vehicles IP addresses social
2+ sources (18+ years old) addresses handles/flags
Data Enrichment | Phone Append | Reverse Phone Append | E-append | Reverse E-append | eCOA | Onboarding | Market Sizing | Lists / Audience
Applications Selection | Segmentation | Predictive Modeling | Syndication | Real-time Marketing
D2C| SEGMENTATION,
TARGETING
Summary of Mobile Insights Data
The Mobile Insights study provides in-depth understanding on mobile brands, usage, attitudes
Retail
Device
Brand Channel & Customer Churn &
Demos Level Usage Satisfaction
Performance Purchase Service Switching
Analysis
Drivers
• Awareness • Location of sign-up • Brand & model • Minutes • # of contacts in • Overall

• Age • Likelihood to switch
• Consideration • Reasons for visiting • OEM and Operating • Spend the last year • Network Quality
• HH Income • Have you switched
channel System consideration • When was the • Network coverage
• Race Ethnicity • Recommendation • Data application • Previous carrier
• Channel • Satisfaction with most recent • Price paid
• Willingness to pay a usage (e.g., SMS, • Reasons for
• Hispanic • Auto. system or
premium information sources phone social media, • Customer Service switching
Acculturation person
• Browsing and • Features used & streaming • Quality of additional
• Education • Type of Plan • Plan level switching
purchase activities desired audio/video, • Reason for
including prepaid, services • Drivers to switch
• Employment mobile banking, calling
All-You-Can-Eat, • Reasons for • Screen Size • Speed and Reliability
• Credit Proxy etc.) • # of calls to • Active likely
postpaid, voice and choosing device • Likelihood to upgrade of mobile data
• Adults/Children in data plans
• Business vs. resolve issue switchers (vs.
• Device purchase • OEM and Operating transmission
personal use • Time on hold Passive)
Household • Perception of Best channel System loyalty • NFC and Barcode • Retention factors
• Marital status • OEM Brands • Sat with CS
• Smartphone Technology • Profile of switchers
• Gender considered penetration • Disconnects and
• Tablets, and MiFi, • Carrier brands
• Purchase drivers • Device feature Transfers
Data card and considered
• Reasons for capabilities and • Issue resolution
Mobile Hotspots
choosing carrier usage • Live Chat
• Carrier brands • OEM and Operating utilization and
considered System Next Device satisfaction
• Satisfaction with
retail experience
D2C| SEGMENTATION,
TARGETING
Rule-based Demographic Segmentation: Detailed Profiles
Segment 1: Segment 2: Segment 3: Segment 4: Segment 5: Segment 6: Segment 7: Segment 8: Segment 9: Segment 10: Industry
Millennials < Millennials > GenX < $75K GenX < $75K GenX > $75K GenX > $75K Boomers+ < Boomers+ < Boomers+ > Boomers+ > Base
Source: Nielsen Mobile Insights. Q3 2015 to Q3 2016 $75K $75K no kids with kids no kids with kids $75K no kids $75K w/ kids $75K no kids $75K w/ kids
Gender Male 38% 50% 46% 36% 55% 57% 52% 41% 69% 66% 50%
Female 62% 50% 54% 64% 45% 43% 48% 59% 31% 34% 50%
18-24 43% 33% - - - - - - - - 12%
25-34 57% 67% - - - - - - - - 18%
Age 35-44 - - 33% 61% 31% 60% - - - - 17%
45-54 - - 67% 39% 69% 40% - - - - 18%
55+ - - - - - - 100% 100% 100% 100% 36%
White 51% 60% 65% 54% 72% 67% 77% 56% 85% 72% 66%
Ethnicity Hispanic 21% 17% 13% 24% 10% 16% 10% 27% 4% 11% 15%
Black/African American 17% 9% 15% 15% 8% 8% 9% 14% 4% 9% 11%
Asian/Pacific Islander 6% 11% 3% 3% 7% 7% 1% 1% 3% 5% 5%
< $50k 72% - 68% 63% - - 66% 70% - - 41%
Income 50k-75k 28% - 32% 37% - - 34% 30% - - 19%
75k-100k - 40% - - 34% 34% - - 33% 35% 14%
100k + - 60% - - 66% 66% - - 67% 65% 26%
Low 28% 16% 26% 22% 11% 10% 17% 18% 7% 10% 18%
Credit Status Medium 54% 49% 54% 61% 42% 49% 48% 63% 27% 47% 48%
High 17% 36% 20% 17% 47% 41% 35% 19% 66% 44% 34%
Employment Full-time 38% 61% 44% 49% 74% 76% 15% 23% 32% 53% 42%
Urban 20% 29% 19% 18% 22% 20% 15% 22% 16% 20% 19%
Location Suburban 50% 51% 48% 50% 53% 54% 48% 46% 56% 55% 51%
Small Town / Rural 29% 20% 33% 33% 25% 26% 36% 32% 28% 25% 30%
Education Ass/BA/BS degree or greater 36% 66% 42% 41% 72% 78% 45% 34% 78% 72% 54%
Phone Type Smartphone penetration 95% 95% 81% 90% 91% 95% 62% 73% 79% 85% 84%
Handset Brand Apple share of smartphone base 37% 51% 24% 29% 46% 49% 20% 20% 41% 41% 35%
Samsung share of smartphone base 31% 26% 33% 35% 29% 29% 31% 33% 29% 30% 31%
Mobile Substitution HH Percent using only wireless service 67% 58% 55% 57% 48% 47% 34% 40% 23% 28% 48%
Price/promotion/cost 30% 20% 34% 33% 26% 23% 33% 35% 22% 25% 28%
Reason for Current Carrier
Voice network quality 9% 12% 9% 9% 15% 15% 12% 8% 19% 12% 12%
Selection
Data network quality 10% 15% 6% 7% 10% 11% 3% 5% 5% 8% 8%
Family/shared Plan Percent on family/shared postpaid plan 71% 71% 58% 75% 70% 75% 60% 72% 76% 79% 70%
HH Postpaid Lines Mean number of postpaid lines Q316 2.94 3.29 2.31 2.94 2.68 3.14 2.19 2.75 2.56 3.18 2.76
HH Postpaid Spend Mean monthly postpaid spend Q316 $142 $149 $137 $162 $159 $172 $108 $147 $144 $171 $144
TMO Share of Base TMO share of segment base 15% 13% 11% 14% 8% 10% 7% 11% 5% 9% 11%
Share of T-Mobile Base Segment share of Magenta base 29% 12% 9% 13% 4% 10% 12% 2% 7% 1% 100%
Share of TMO GAs Segment share of Magenta GAs Q316 36% 20% 6% 13% 2% 10% 6% 3% 3% 1% 100%
D2C| SEGMENTATION,
TARGETING
Mapping Target Segments to Nielsen ConneXions Segments
Segments Magenta Primary (Media Target) Magenta Secondary/Metro Overlap Magenta Stretch Magenta Non-Target
Big City, Small Tech (28)
IM Nation (29)
Seg 1 w/o kids Techs and the City (30)
Mapping based on highest occurring index of CX segment within target segment base vs total market base
Gearing Up (34)
Seg 1.5 w/kids Calling Circles (8)
You & I Tunes (5)
Seg 2 w/o kids Time Shifters (17)
Generation WiFi (7)
Seg 2.5 w/kids Tech Nests (3)
Plug & Play (31) Digital Dreamers (33)
Opting Out (36) Techtown Lites (37)

Seg 3 w/o kids
New Kids on the Grid
Video Homebodies (39)
(38)
Cyber Strivers (23) Dish Country (9)

Seg 4 w/kids
Video Vistas (27) Family Dishes (32)
WiFi Warriors (11) The Pragmatics (14)

Seg 5 w/o kids Cyber Sophisticates (13) Satellites & Silos (12)
Cinemaniacs (20) New Technorati (18)
Plugged-In Families (2)
Seg 6 w/kids Bundled Burbs (15) Technovators (1) Connected Country (4)
Kids & Keyboards (16)
Bucolic Basics (43) Discounts & Deals (47)
Dial-Up Duos (48) Satellite Seniors (49)

Seg 7 w/o kids Antenna Land (41)
Mapping based on highest occurring index of CX segment within target segment base vs total marketEarly-Bird
base TV (50) Tech-Free Frontier (51)
The Unconnected (52) Last to Adopt (53)
Multimedia Families (21) Rural Transmissions (26) Low-Tech Country (40)

Seg 8 w/kids
Leisurely Adopters (44) Tech Skeptics (42) Landline Living (45)
Internet Hinterlands
Analoggers (22) (24)
Seg 9 w/o kids Broadband Boulevards (35)
Low-Speed Boomers Old-Time Media (46)
(25)
High-Tech Society (6)
Seg 10 w/kids Smart Gamers (10)
D2C| SEGMENTATION,
TARGETING
T-Mobile Confidential Gadgets Galore (19) 48 48
TMO Demographic Segments By ConneXions
Consumer Segment by ConneXions
Millennials < Millennials > GenX < $75K GenX < $75K GenX > $75K GenX > $75K Boomers+ < Boomers+ < Boomers+ > Boomers+ >
Code ConneXions Segment LG Lifestage Group Demographic Caption Techno Class Magenta People Classification $75K $75K no kids with kids no kids with kids $75K no kids $75K w/ kids $75K no kids $75K w/ kids
1 Technovators F1 Early-Adopting Elite Wealthy Younger w/ Kids High-Tech Magenta Stretch 0.8% 3.0% 0.4% 0.7% 2.6% 6.3% 0.3% 0.3% 1.0% 2.0%
2 Plugged-In Families F1 Early-Adopting Elite Upscale Younger w/ Kids High-Tech Magenta Primary 1.4% 3.7% 1.0% 1.9% 3.7% 8.4% 0.5% 0.9% 1.3% 2.0%
3 Tech Nests F1 Early-Adopting Elite Midscale Younger w/ Kids High-Tech Magenta Primary 2.9% 3.5% 0.9% 2.7% 1.4% 2.8% 0.6% 1.8% 0.5% 1.4%
4 Connected Country F1 Early-Adopting Elite Upscale Middle Age w/ Kids High-Tech Non-Target 1.0% 1.9% 0.6% 1.0% 2.1% 3.8% 0.5% 0.8% 1.4% 3.1%
5 You & I Tunes Y1 Young & Wireless Upper Mid Younger w/o Kids High-Tech Magenta Primary 2.9% 5.3% 1.2% 0.9% 1.8% 1.7% 0.8% 0.9% 1.0% 1.3%
6 High-Tech Society F1 Early-Adopting Elite Wealthy Older w/ Kids High-Tech Magenta Stretch 0.7% 3.1% 0.4% 0.3% 3.3% 3.1% 0.4% 0.2% 3.4% 6.1%
7 Generation WiFi Y1 Young & Wireless Midscale Younger w/o Kids High-Tech Magenta Primary 3.3% 5.4% 1.5% 1.3% 2.0% 1.6% 0.6% 0.8% 0.6% 1.3%
8 Calling Circles F2 Suburban Spenders Midscale Younger w/ Kids High-Tech Secondary Magenta 2.4% 2.0% 1.4% 2.3% 1.1% 1.8% 0.6% 1.4% 0.3% 1.1%
9 Dish Country F2 Suburban Spenders Midscale Younger w/ Kids High-Tech Non-Target 1.8% 1.1% 1.0% 2.9% 0.9% 1.7% 0.5% 0.9% 0.3% 0.8%
10 Smart Gamers F2 Suburban Spenders Upscale Older w/ Kids High-Tech Magenta Stretch 1.0% 2.7% 0.7% 0.7% 4.4% 3.1% 0.5% 1.0% 3.4% 7.9%
11 WiFi Warriors M1 Wireless White-Collars Wealthy Middle Age w/o Kids High-Tech Magenta Primary 1.0% 2.5% 1.0% 0.8% 4.5% 2.8% 0.6% 0.5% 2.2% 1.6%
12 Satellites & Silos Y1 Young & Wireless Upper Mid Middle Age w/o Kids High-Tech Non-Target 1.8% 1.8% 1.6% 1.5% 2.9% 1.7% 1.0% 1.3% 2.1% 1.6%
13 Cyber Sophisticates M1 Wireless White-Collars Wealthy Middle Age w/o Kids High-Tech Magenta Stretch 0.8% 2.7% 0.8% 0.5% 4.7% 3.1% 0.7% 0.5% 3.3% 2.5%
14 The Pragmatics Y1 Young & Wireless Upscale Middle Age w/o Kids High-Tech Magenta Primary 1.4% 2.6% 1.4% 1.4% 3.8% 2.5% 0.7% 0.8% 1.3% 1.6%
15 Bundled Burbs F2 Suburban Spenders Upscale Middle Age w/ Kids High-Tech Magenta Primary 1.0% 2.0% 0.9% 1.4% 3.5% 5.1% 0.5% 0.7% 1.2% 3.3%
16 Kids & Keyboards F2 Suburban Spenders Upper Mid Middle Age w/ Kids High-Tech Magenta Primary 2.3% 2.7% 2.2% 5.2% 3.1% 6.7% 0.9% 2.3% 0.8% 2.7%
17 Time Shifters Y1 Young & Wireless Upper Mid Younger w/o Kids High-Tech Magenta Primary 1.7% 2.1% 1.7% 1.7% 2.3% 1.9% 0.7% 0.8% 0.8% 0.8%
18 New Technorati Y2 Emerging Techies Upper Mid Older w/o Kids Mid-Tech Magenta Primary 2.3% 2.6% 2.7% 2.3% 2.6% 2.0% 1.1% 1.5% 1.0% 1.1%
19 Gadgets Galore F3 Frugal Folks Upper Mid Older w/ Kids Mid-Tech Magenta Stretch 1.5% 1.9% 1.5% 1.7% 2.5% 2.2% 2.2% 3.5% 4.1% 7.4%
20 Cinemaniacs M2 Mature Mid-Techs Upscale Older w/o Kids Mid-Tech Magenta Primary 1.7% 2.3% 2.6% 1.5% 4.9% 2.0% 1.7% 1.3% 3.3% 2.9%
21 Multimedia Families F3 Frugal Folks Upper Mid Older w/ Kids Mid-Tech Secondary Magenta 2.3% 1.7% 3.0% 4.1% 1.5% 2.1% 1.8% 5.0% 0.7% 2.4%
22 Analoggers M2 Mature Mid-Techs Wealthy Older w/o Kids Mid-Tech Non-Target 0.8% 2.3% 0.7% 0.5% 2.7% 1.9% 1.2% 0.3% 10.9% 3.2%
23 Cyber Strivers F3 Frugal Folks Lower Mid Middle Age w/ Kids Mid-Tech Secondary Magenta 2.1% 0.9% 1.6% 2.6% 0.5% 0.7% 0.6% 2.0% 0.2% 0.3%
24 Internet Hinterlands M2 Mature Mid-Techs Upscale Older Mostly w/o Kids Mid-Tech Non-Target 0.8% 0.8% 1.0% 0.8% 1.5% 0.7% 2.1% 1.3% 4.8% 2.4%
25 Low-Speed Boomers M2 Mature Mid-Techs Wealthy Older w/o Kids Mid-Tech Non-Target 1.0% 2.1% 1.0% 0.7% 2.9% 1.8% 2.1% 0.8% 10.1% 4.1%
26 Rural Transmissions M2 Mature Mid-Techs Midscale Older Mostly w/o Kids Mid-Tech Non-Target 1.6% 0.8% 2.8% 1.9% 1.5% 1.0% 2.5% 3.0% 1.1% 1.7%
27 Video Vistas F3 Frugal Folks Lower Mid Middle Age w/ Kids Mid-Tech Secondary Magenta 2.5% 1.2% 2.3% 3.8% 0.8% 1.4% 0.9% 2.5% 0.2% 1.0%
28 Big City, Small Tech F3 Frugal Folks Low Income Younger w/ Kids Mid-Tech Secondary Magenta 2.8% 1.1% 1.6% 3.0% 0.3% 0.5% 0.7% 2.1% 0.1% 0.4%
29 IM Nation Y2 Emerging Techies Lower Mid Younger w/o Kids Mid-Tech Secondary Magenta 3.4% 1.9% 1.8% 2.0% 0.7% 1.0% 1.0% 1.4% 0.3% 0.8%
30 Techs and the City Y2 Emerging Techies Lower Mid Younger w/o Kids Mid-Tech Secondary Magenta 4.3% 3.2% 2.6% 2.5% 1.5% 1.4% 1.4% 1.9% 0.5% 0.5%
31 Plug & Play Y2 Emerging Techies Upper Mid Older w/o Kids Mid-Tech Secondary Magenta 2.1% 2.1% 3.1% 2.3% 2.7% 1.7% 1.2% 1.5% 0.7% 1.0%
32 Family Dishes F3 Frugal Folks Lower Mid Middle Age Family Mix Mid-Tech Non-Target 3.3% 1.2% 3.8% 4.9% 0.8% 1.1% 1.8% 3.7% 0.4% 0.9%
33 Digital Dreamers Y2 Emerging Techies Lower Mid Older w/o Kids Mid-Tech Secondary Magenta 3.8% 2.2% 4.3% 4.3% 1.7% 1.9% 1.9% 2.9% 0.7% 1.4%
34 Gearing Up Y2 Emerging Techies Low Income Middle Age w/o Kids Mid-Tech Secondary Magenta 3.2% 1.6% 1.9% 1.6% 0.4% 0.6% 0.8% 1.0% 0.2% 0.2%
35 Broadband Boulevards M2 Mature Mid-Techs Upper Mid Older w/o Kids Mid-Tech Magenta Stretch 1.4% 2.4% 1.8% 1.2% 2.5% 1.8% 3.2% 2.0% 5.8% 3.7%
36 Opting Out M2 Mature Mid-Techs Upper Mid Older w/o Kids Mid-Tech Secondary Magenta 1.7% 1.3% 2.9% 1.4% 2.1% 1.2% 2.6% 2.1% 2.4% 2.0%
37 Techtown Lites Y3 Young Startups Lower Mid Older Mostly w/o Kids Low-Tech Secondary Magenta 2.3% 1.0% 3.3% 2.7% 0.8% 0.6% 1.6% 2.9% 0.4% 1.2%
38 New Kids on the Grid Y3 Young Startups Downscale Older Mostly w/o Kids Low-Tech Secondary Magenta 1.7% 1.0% 2.0% 1.4% 0.7% 0.5% 1.0% 1.4% 0.2% 0.3%
39 Video Homebodies F3 Frugal Folks Low Income Older Mostly w/o Kids Mid-Tech Secondary Magenta 2.1% 0.9% 2.2% 2.4% 0.4% 0.5% 0.9% 2.3% 0.2% 0.7%
40 Low-Tech Country Y3 Young Startups Downscale Older Mostly w/o Kids Low-Tech Non-Target 1.3% 0.4% 2.2% 1.3% 0.4% 0.3% 1.8% 3.1% 0.2% 0.3%
41 Antenna Land M3 Offline Seniors Midscale Older w/o Kids Low-Tech Secondary Magenta 3.6% 2.3% 4.5% 3.4% 2.6% 2.1% 7.7% 6.0% 2.7% 3.6%
42 Tech Skeptics M3 Offline Seniors Upper Mid Mature w/o Kids Low-Tech Non-Target 1.1% 1.5% 1.5% 1.0% 2.0% 1.4% 4.9% 2.2% 7.2% 3.0%
43 Bucolic Basics M3 Offline Seniors Midscale Mature w/o Kids Low-Tech Non-Target 1.0% 0.5% 1.5% 1.0% 1.0% 0.6% 4.3% 2.0% 2.2% 1.9%
44 Leisurely Adopters M3 Offline Seniors Low Income Older w/o Kids Low-Tech Secondary Magenta 2.4% 1.3% 2.9% 2.2% 0.6% 0.9% 2.7% 3.2% 0.3% 0.6%
45 Landline Living M3 Offline Seniors Downscale Older Mostly w/o Kids Low-Tech Non-Target 1.7% 0.5% 2.0% 1.7% 0.4% 0.5% 2.5% 2.7% 0.3% 0.4%
46 Old-Time Media M3 Offline Seniors Upper Mid Mature w/o Kids Low-Tech Non-Target 0.8% 1.3% 1.4% 0.8% 1.9% 1.4% 3.3% 1.5% 5.0% 2.1%
47 Discounts & Deals M3 Offline Seniors Downscale Older Mostly w/o Kids Low-Tech Non-Target 1.4% 0.8% 1.6% 1.3% 0.6% 0.5% 2.3% 2.4% 0.7% 0.9%
48 Dial-Up Duos M3 Offline Seniors Midscale Mature Mostly w/o Kids Low-Tech Non-Target 1.6% 1.3% 2.6% 1.9% 1.7% 1.2% 7.2% 3.8% 2.7% 2.4%
49 Satellite Seniors M3 Offline Seniors Downscale Older Mostly w/o Kids Low-Tech Non-Target 0.8% 0.2% 1.0% 1.0% 0.2% 0.3% 2.1% 1.8% 0.3% 0.2%
50 Early-Bird TV M4 Elderly Traditionalists Midscale Mature w/o Kids No-Tech Non-Target 0.9% 1.0% 1.4% 0.9% 1.2% 0.8% 3.9% 0.9% 2.0% 1.4%
51 Tech-Free Frontier M4 Elderly Traditionalists Downscale Mature w/o Kids No-Tech Non-Target 1.2% 0.5% 1.8% 1.4% 0.5% 0.4% 3.5% 1.6% 0.8% 0.3%
52 The Unconnected M4 Elderly Traditionalists Downscale Mature w/o Kids No-Tech Non-Target 2.3% 1.6% 3.3% 2.5% 1.6% 1.4% 5.5% 3.6% 1.5% 1.5%
53 Last to Adopt M4 Elderly Traditionalists Low Income Older w/o Kids No-Tech Non-Target 2.4% 1.3% 2.9% 2.3% 0.8% 1.0% 3.8% 2.9% 0.4% 0.6%
99 Unclassified 99 Unassigned Unassigned Unassigned Non-Target 0.6% 0.6% 0.3% 0.4% 0.2% 0.4% 0.1% 0.1% 0.1% 0.2%
100% 100% 100% 100% 100% 100% 100% 100% 100% 100%
D2C| SEGMENTATION,
TARGETING
Rule-based Segmentation and Promos: Base Data
 T-Mobile

Millennials <
$75K
Millennials >
$75K
GenX < $75K
no kids
GenX < $75K
with kids
GenX > $75K
no kids
GenX > $75K
with kids
Boomers+ <
$75K no kids
Boomers+ <
$75K w/ kids
Boomers+ >
$75K no kids
Boomers+ >
$75K w/ kids Total
Instant POS rebate 5.7% 12.1% 5.2% 8.1% 6.4% 6.8% 3.5% 6.8% 4.1% 5.4% 6.8%
Service bundling 6.3% 11.0% 2.0% 4.7% 3.1% 6.7% 2.0% 1.8% 1.8% 0.8% 5.6%
Device discounts 8.7% 13.2% 6.7% 9.3% 7.9% 9.4% 5.2% 8.6% 5.6% 4.9% 8.8%
Free Device BOGO 5.6% 9.0% 2.6% 6.0% 3.0% 8.2% 1.8% 2.2% 1.5% 7.5% 5.5%
$0 down for new device 17.4% 17.2% 17.8% 21.0% 13.7% 16.9% 13.1% 17.6% 9.0% 11.3% 16.8%
Mail-in rebate 6.0% 8.3% 3.6% 4.2% 7.4% 7.3% 2.5% 0.2% 4.8% 1.7% 5.6%
Bundled accessories 8.0% 14.3% 5.6% 6.9% 8.1% 10.6% 3.1% 4.8% 4.4% 9.2% 8.2%
Device trade-in special 11.9% 15.2% 8.8% 10.9% 10.2% 13.3% 7.5% 9.9% 11.7% 16.1% 11.7%
Switching credit 4.2% 6.7% 2.4% 4.1% 4.2% 4.0% 1.1% 2.2% 2.0% 4.3% 4.0%
Online specials 5.8% 11.4% 2.6% 4.1% 3.3% 8.7% 1.1% 3.5% 0.7% 2.3% 5.6%
Value added promos 6.0% 10.3% 1.8% 3.8% 4.2% 8.2% 0.9% 2.3% 1.3% 1.5% 5.4%
Gift card offering 4.8% 8.0% 0.9% 2.6% 3.8% 6.9% 1.1% 0.4% 1.5% 3.5% 4.2%
Additional airtime 3.7% 9.3% 1.7% 2.4% 2.6% 6.2% 1.8% 1.8% 1.8% 3.3% 4.1%
Other promotion 3.0% 1.9% 3.8% 3.9% 3.4% 3.0% 8.1% 3.1% 6.8% 1.2% 3.7%
Participated in no promo 45.5% 33.5% 57.3% 46.2% 53.7% 42.6% 64.0% 49.3% 64.2% 52.2% 47.5%
 Non-T-Mobile

Millennials <
$75K
Millennials >
$75K
GenX < $75K
no kids
GenX < $75K
with kids
GenX > $75K
no kids
GenX > $75K
with kids
Boomers+ <
$75K no kids
Boomers+ <
$75K w/ kids
Boomers+ >
$75K no kids
Boomers+ >
$75K w/ kids Total
Instant POS rebate 8.5% 11.7% 9.3% 11.2% 8.5% 11.9% 7.5% 10.2% 10.3% 9.6% 9.7%
Service bundling 5.5% 9.9% 2.7% 4.2% 3.0% 7.6% 2.1% 3.1% 2.6% 6.6% 4.9%
Device discounts 13.5% 16.9% 11.1% 12.8% 15.7% 16.8% 10.3% 12.0% 14.2% 14.3% 13.7%
Free Device BOGO 5.9% 8.7% 3.1% 5.8% 3.5% 7.2% 2.3% 4.4% 2.3% 3.2% 5.0%
$0 down for new device 11.9% 12.9% 10.8% 12.0% 10.3% 13.0% 8.2% 12.3% 9.3% 12.7% 11.2%
Mail-in rebate 5.4% 8.6% 3.5% 5.2% 4.2% 7.0% 3.1% 4.5% 4.8% 4.3% 5.3%
Bundled accessories 7.6% 12.7% 4.3% 6.4% 5.9% 9.7% 3.6% 4.8% 6.3% 6.4% 7.1%
Device trade-in special 10.9% 16.1% 8.5% 9.8% 11.0% 14.1% 7.8% 9.2% 12.3% 11.0% 11.2%
Switching credit 3.7% 5.9% 1.6% 2.9% 1.4% 4.5% 1.0% 1.9% 0.9% 1.8% 2.9%
Online specials 4.9% 7.9% 2.6% 4.1% 3.2% 5.8% 1.8% 2.0% 1.1% 2.1% 4.0%
Value added promos 3.7% 7.5% 1.5% 3.2% 1.7% 5.2% 0.6% 1.2% 0.9% 1.5% 3.1%
Gift card offering 3.9% 8.6% 2.4% 4.0% 2.3% 6.8% 1.4% 2.3% 2.4% 2.4% 4.0%
Additional airtime 3.4% 7.6% 2.3% 3.4% 2.0% 6.0% 3.2% 2.5% 1.9% 3.7% 3.8%
Other promotion 3.5% 1.9% 6.7% 4.7% 4.8% 3.5% 8.1% 8.2% 6.0% 5.7% 4.8%
Participated in no promo 45.8% 37.0% 52.5% 44.5% 49.5% 38.7% 55.7% 50.8% 50.4% 47.7% 46.7%
D2C| SEGMENTATION,
TARGETING
Inadequacy of Existing Targeting Solutions
Assessed Targeting Options

Proposal
Machine-
Carrier- Channel- Develop a deep
Application Demographic based
based based behavioral
What is the approach? Classification Propensity
Schema Segments
Model segmentation
Build generic Use diverse Rely only on
Retain DTC’s
target profiles ad hoc, automated, Hook to Claritas
Implication existing
differentiated channel- machine-
What does it involve? demographic ConneXions
only by specific learning
segments
carrier segmentation model
Build Magenta
Ends in poor
Is insufficiently Often yield
Is web-specific; People construct
predictive; highly unstable
target delineation, excludes vast
implicitly segments; lack
Complication thus will severely
assumes cohort cross-channel
array of
What are the key drawbacks? limit offer and behavioral and Map to machine
homogeneity; has validity; offer
message psychographic
low web match very thin target learning model
differentiation inputs
rate profiling
D2C| SEGMENTATION,
TARGETING

Segmentation Roadmap Sept 5 2017

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Segmentation Roadmap Sept 5 2017

Uploaded by

Copyright:

Available Formats

Prospect Segmentation

D2C mission is to smartly leverage

Low/Start Delivery Timeframe End

(1) Q3 - Q4 2016 (2) Q4 2016 (3) Q1 2018 (4) Q2 2018

“Purple Land” “Common Market”

“New Frontier” “Magenta Land”

Aspirational Magenta Common Purple

Smart Multimedia Big City, Landline

Target as Magenta if mid-scale or higher

Nielsen ConneXions Comparison

41% Use smartphone for fantasy sports,

95% own smartphone and 48% own tablet;

Total GA Segment Movement If referred to simply as “gross adds” you can

Source: Nielsen Mobile Insights, blended postpaid and prepaid

T-Mobile Segments - Recent Promo Participation

Non-T-Mobile Segments - Recent Promo Participation

Product Features by Segment: TMO vs. Market

Index of TMO to Market by Segment

Boomers 55+ Prospects

Segments 7/9 Segments 8/10

Segments 8/10 <= 2

Group A: Group B: Group C: Group D:

target variable for segmentation

drawn from TMO demographics model is

Voice network Data network Monthly Overall

Inputs* Method Deliverables

Credit card purchases Individual Merkle ●   Expected Merkle module

Customer web behavior Individual Engineering ●  ◕ Dependency on Engineering

Prospect tech behavior Individual Claritas ◐  ◔ Requires append

Data Analysis Engines

Channel Key Secondary

Magenta Core (Media) Magenta Core (Non Media) Magenta Stretch

DEMOGRAPHICS LIFESTYLE ATTITUDE TECHNOLOGY MOBILE SOCIAL MEDIA ONLINE

 Affinity model inputs will be determined through testing and optimization

Step 1 Input Processing/Evaluation TBD Leadership approval September 2017

Step 2 Data Migration to Jolt TBD MET workload October 2017

Step 3 Segmentation Dataset Prep TBD Jolt readiness November 2017

Step 4 Behavioral Segmentation Build TBD Jolt readiness Jan-Mar 2018

Step 6 Testing and Deployment TBD Segment completion June-Sep 2018

 Sample data processing

129MM 275MM 500MM 475MM 130MM 291MM 75MM

• Awareness • Location of sign-up • Brand & model • Minutes • # of contacts in • Overall

Seg 1 w/o kids Techs and the City (30)

Seg 1.5 w/kids Calling Circles (8)

You & I Tunes (5)

Seg 2 w/o kids Time Shifters (17)

Generation WiFi (7)

Seg 2.5 w/kids Tech Nests (3)

Plug & Play (31) Digital Dreamers (33)

Opting Out (36) Techtown Lites (37)

Cyber Strivers (23) Dish Country (9)

WiFi Warriors (11) The Pragmatics (14)

Plugged-In Families (2)

Kids & Keyboards (16)

Bucolic Basics (43) Discounts & Deals (47)

Dial-Up Duos (48) Satellite Seniors (49)

The Unconnected (52) Last to Adopt (53)

Multimedia Families (21) Rural Transmissions (26) Low-Tech Country (40)

High-Tech Society (6)

Seg 10 w/kids Smart Gamers (10)

Source: Nielsen Mobile Insights, Q4 2015 to Q4 2016

Assessed Targeting Options

You might also like