Wireless Telecom is a complex category, with multiple ad messaging from ads for cellphones, cellphone plans and branded messages claiming network superiority. This is an actual case study that sorts all of this out and illustrates the power of marketing measurement and ROI assessment
2. • The following is a real case study from a
telecom industry. Data, labels and brand
names have been masked to preserve
confidentiality.
Overview and Disclaimer
3. Outline
Model architecture
Background
Key Insights
Decomposing New Subscribers YE July 07: national, by region and
across time.
New Subscribers variance drivers: drivers of New Subscribers
trends YE July 07.
Relative sensitivities across media message mix: national and by
region
The relative effectiveness of TV media messaging: national and by
region
The relative impact of message executions and clutter: national and
by region
The impact of media quality and persuasion: national and by region.
The impact of customer satisfaction on new subscribers
Marketing Return on Investment
4. Outline
Optimizing total Marketing Spending
Optimizing the marketing spending mix, national
Optimizing local marketing spending
Conclusions and recommendations
Model validations
5. Model Architecture
Promotion
By Individual commercial & message groupNational TV GRPs
New
Subscribers
Jan05-Jul07
By week
Region (4)
x DMA(50)
Competitor GRPs
Competitors A, B and C
Copy Quality Copy Test x GRPs weighted
Local Radio GRPs by DMA and region
Local OOH
GRPs by DMA and region
Local Spot
TV
GRPs by DMA and region
Online Adv GRPs by week
Local Product Promo
Print ad spend by region/DMA: Price-off, Rebates,
BOGOs and estimated Price Discount levels.
National Promo
Print ad promo spend natl pubs WSJ, USA Today. etc.
Customer Sat.
Customer satisfaction by DMA/Region
Christmas, NCAA promo
Seasonality
Message Print Adv.
National Print GRPs by media message
By Month
The most comprehensive model
to-date includes drivers covering
national TV, local print and
secondary media, handset
promotional advertising and
discounts, competitive GRPs and
Customer Satisfaction pooled by
DMA and region.
6. Background
The client is a key player in a solid growth segment of the telecom
industry. There are some unique aspects of this client and
category.
The client produces and airs about 65 different TV commercial
messages per year. At times, as many as 15 different commercial
messages are aired in a single week. Because of this, we actually
include the frequency or number of commercial executions per
week, as the negative effect of too many messages (clutter) is a
key issue.
The client and industry is a very heavy user of national TV
advertising and such advertising is a very significant factor in
driving new subscribers. Because of the heavy amount of
advertising and promotion, it is critical to understand the
effectiveness of each media and message.
Because of the large number of messages communicated to
consumers, our efforts will also focus on understanding and
quantifying the effectiveness of different media messages.
7. Background
The company’s media strategy centers around an architecture of
different marketing messages. These messages include 1)
branded messages, 2) product messages, 3) plans and pricing
messages, 4) service messages and 5) messages related to a
major new product launch during the year.
Because of the large number of messages communicated to
consumers, our efforts will also focus on understanding and
quantifying the effectiveness of different media messages.
Because or modeling approach focuses down to the individual
commercial level, we are equipped to do this
8. Key Insights
For the quarter ending July ’07, Company’s New Subscribers
performance recovered significantly from the soft prior quarter, with a
period-to-period gain of +19.3%.
The major news for the period includes a significant “lift” from the
launch of a new smart-phone. This phone contributed a 4.4 percent
incremental sales for the quarter and generated a lift of +23% for the
first five weeks after it’s introduction in the last week of June
While the smart-phone was a significant contributor to growth for the
period, Company’s media messaging generated about an equal
positive impact. For the period, Company moved towards a more
optimal allocation of media with lower execution frequencies for non-
branded messaging and a significant increase in branded message
media.
When analyzing media and marketing effects across regions, we
continue to see a high degree of diverse effects. Each region has
their particular marketing strengths and marketing-mix effects. The
NE region was found to be the most impacted by competitive media.
The SE region was most affected by TV media, while the West region
was most impacted by secondary and radio media; and the Central
region was most impacted by handset promotional print ads.
9. Key Insights
While Company improved considerably in the efficiency of their media
messaging over the past quarter, considerable upside continues to exist by
moving towards an optimal mix of media and marketing.
Company’s marketing spend is profitable, earning a 71% annual return on
spending of about $900MM Given current marketing response, and
assuming the current spending mix, positive marginal returns should be
expected up to 2X current overall spending levels.
When considering optimization of the national media and marketing mix, the
plan calls for higher GRPs for both branded and non-branded messaging,
plus increases for product promotion and new product advertising. To
balance the constant spending optimization, reductions in spending are
indicated for print advertising, radio, OOH advertising, online and Spot TV.
This solution is expected to net an additional +5.9% in new subscribers,
When considering optimization of local media, the plan indicates the greatest
upside opportunity by shifting spending away from the SE and West regions
and increasing spending in the NE and Central regions. This solution will
add an additional +1.4% in new subscribers.
Overall, by implementing our optimal spending plan, we estimate Company
can increase New Subscribers by +7.3% at constant spending.
10. Company National Decomposition of New Subscribers Year-Ending Ending July 07
34.1%
6.6%
9.5%
6.4%
4.2%
3.2%
7%
1.2%
1.8%
4.6%
2.0%
4.4%
6.9%
2.4%
65.9%
BASE
CUSTOMER SAT
BRANDED.EXECUTIONS
NONBRANDED.EXECUTIONS
TV.BRANDED MESSAGE 1
TV.BRANDED MESSAGE 2
TV.BRANDED MESSAGE 3
TV.BRANDED MESSAGE 4
TV.TELECOM PLANS & PRICING MESSAGES
TV.TELECOM PRODUCT MESSAGE
TV.TELECOME SERVICES MESSAGE
TV.NEW PRODUCT LAUNCH MESSAGE
TV.COPY QUALITY
RADIO
OOH
SPOTTV
ONLINE
PRINT
PRODUCT PROMOS
PROMOTION & SPONSORSHIP
For the year ending July ‘07, marketing drove 59.5% of New Subscribers. Consistent with
past models, branded message TV was the largest contributor at 20.4%, followed by
non-branded message TV at 5.8%. Customer satisfaction was found to be a significant
driver at 6.6%. Message print drove 6.9% of New Subscribers, while product promo-
tional print drove 5.5% of New Subscribers and competitive media had a -6.8% impact
11. Company Regional Decomposition of New Subscribers Q107
-20%
0%
20%
40%
60%
80%
100%
NE SE Central West National
COMPETITIVE PRINT
COMPETITOR B TV
COMPETITOR A TV
PROMOTION &
SPONSORSHIP
NEW PRODUCT LAUNCH
MESSAGES
NONBRANDED.TV.MESSAGE
BRANDED.TV.MESSAGE
COPY QUALITY
HANDSET.PROMO
PRINT
ONLINE
SPOTTV
OOH
RADIO
NONBRANDED.EXECUTIONS
BRANDED.EXECUTIONS
CUSTOMER SATISFACTION
When decomposing marketing impacts by region, we see a significant degree of heterogeneity.
The NE region is significantly more impacted marketing drivers than the other
Regions and also was more impacted by competitive media, especially Competitor A. The
SE region had the largest impact from message TV and the Central region was most
Impacted by product promotion.
12. Decomposition of Company New Subscribers by Week
-100,000
0
100,000
200,000
300,000
400,000
500,000
600,000
01.02.05
02.06.05
03.13.05
04.17.05
05.22.05
06.26.05
07.31.05
09.04.05
10.09.05
11.13.05
12.18.05
01.22.06
02.26.06
04.02.06
05.07.06
06.11.06
07.16.06
08.20.06
09.24.06
10.29.06
12.03.06
01.07.07
02.11.07
03.18.07
04.22.07
05.27.07
07.01.07
COMPTV.PRINT
COMPETITIOR B TV
COMPETITOR A TV
PROMOTION
NEW PRODUCT LAUNCH TV
NONBRANDED TV MESSAGE
BRANDED TV MESSAGE
COPY QUALITY
PRODUCT PROMOS
PRINT
ONLINE
SPOTTV
OOH
RADIO
NONBRANDED EXECUTIONS
BRANDED.EXECUTIONS
CUSTOMER.SAT
BASE
The Company’s New Subscribers increased +19.3% for quarter ending July v. the prior quarter.
Over the last five weeks of the period, we see a significant impact from the Smart-Phone launch,
with this event driving over a +23% lift on New Subscribers.
13. Quarter Ending July v. Prior Variance Drivers as % of US
-1.5%
-0.9%
-0.6%
-0.3%
-0.1%
-0.1%
-0.0%
0.2%
0.2%
0.3%
0.7%
0.9%
1.0%
1.2%
1.5%
1.6%
3.8%
-4% -2% 0% 2% 4% 6% 8%
CUST.SAT
RADIO
COMPETITOR A MEDIA
PROMOTION
ONLINE
SPOTTV
OOH
COMPETITOR B MEDIA
COMPETITIvE PRINT
PRINT
BRANDED MESSAGE TV
COPY QUALITY
PRODUCT PROMO
NONBRANDED TV EXECUTIIONS
NEW PRODUCT LAUNCH TV
BRANDED TV EXECUTIONS
BASE MOMENTUM
NE
SE
CENTRAL
WEST
From the prior year, New Subscribers increased +7.8%. The SE and West regions
posted the largest relative increases. All regions showed growth over the prior quarter, while the NE
region posted a small decline. Branded TV executions and base momentum were the largest posi-
tive drivers, while radio and declining customer satisfaction were the largest negative drivers.
14. Impact of Media Messaging
235,500
236,000
236,500
237,000
237,500
238,000
238,500
239,000
0 50 100 150 200 250 300 350
Company Mobile Subscribers Sensitivities
to Weekly Message GRPs
TOTAL
NONBRANDED.TV.
MESSAGES
TOTAL.BRANDED.T
V.MESSAGES
TELECOM PLANS &
PRICING
MESSAGES
TELECOM
PRODUCT
MESSAGE
TELECOME
SERVICES
MESSAGE
NEW PRODUCT
LAUNCH MESSAGE
0
5
10
15
20
25
30
35
40
Company Incremental Gross Adds per
Incremental 100 Message GRPs
Northeast
Southeast
Central
West
Branded message TV is slightly over 2x more responsive to increases in GRPs than non-
Branded message TV.. Per incremental GRP of spending, the SE region showed the highest
total increase in New Subscribers per incremental GRP.
15. Media Message Impact per 100 GRPs
0 500 1,000 1,500 2,000
New Subs
per Incr.
100 GRPs
New Product Launch
Message
Telecom Services
Message
Telecom Products
Message
Telecom Plans & Pricing
Message
Branded Message 4
Branded Message 3
Branded Message 2
Branded Message 1
All Non-Branding
Messages
All Branded Messages
When looking at the relative effectiveness of media messaging, we find that branded message
media is about 135% more effective than non-branded messaging. Within branded messaging,
Brand Message 1 showed the highest overall impact.
16. Impact of Media Execution Clutter
Company Sensitivities to Weekly
Commercial Media Executions
195,000
200,000
205,000
210,000
215,000
220,000
225,000
230,000
0 2 4 6 8 10 12
NonBranded Executions
Branded Executions
Company Incremental Subscribers per
Incremental Media Execution
-1000
-500
0
500
1000
1500
2000
Non-Network
Executions
Network Executions
Northeast
Southeast
Central
West
As found in prior modeling exercises, message execution frequency impacts New Subscribers
differently depending on the message, with branded message executions showing a
positive impact, while non-branded executions have a negative impact on New Subscribers.
Across the regions, there is some differences in how executions impact New Subscribers.
While non-branded executions tend to show a negative impact nationally, their effect on
New Subscribers was found to be positive in the West region.
17. Impact of Media Quality
88,000
90,000
92,000
94,000
96,000
98,000
100,000
102,000
104,000
10 15 20 25 30
Company Sensitivities to Copy Test
Persuasion (weekly new subscribers)
Copy Test…
0
100
200
300
400
500
600
Copy Test Persuasion
Company Incremental Subscribers per
Incremental +1 pt. Copy Test Persuasion
Northeast
Southeast
Central
West
By weighting GRPs by copy-test persuasion scores, our model is uniquely able to measure
the impact of qualitative differences in media across the message mix. With average copy
Persuasion scores of 16, Company New Subscribers are about 5% higher than would occur if
all copy scores were at the telecom norm of 12. On an annual basis, this is equivalent to about
$274MM in customer lifetime value.
18. Impact of Customer Satisfaction
Company Sensitivities to Customer
Satisfaction
200,000
210,000
220,000
230,000
240,000
250,000
260,000
0% 20% 40% 60% 80% 100%
FirstCallSat
Company Incremental Subscribers per
Incremental +1% Customer
Satisfaction
0
5
10
15
20
25
Incremental Impact from +1Pt. CP Persuasion
Northeast
Southeast
Central
West
For the first time, our models now incorporate a measure of customer service performance as
a driver of New Subscribers. The measure determined to be most relevant was “First Call Proble
Resolution”. In total, this metric drove a significant 6.6% of overall gross ads in the July quar-
ter. Across the regions, the SE and Central regions showed the greatest impact, while the
NE region was significantly less impacted than the other regions.
19. Marketing ROI
Overall, company makes a solid +71 percent positive return on
$904MM in spending.
Most of net returns are generated through branded TV message
advertising. Other positive return media include product promotion,
non-branded TV and new product messaging. All other media
generated negative net returns.
Source for spending data: MEC
20. Total Marketing Spend and Net Returns
$0
$2,000
$4,000
$6,000
$8,000
$10,000
$12,000
$14,000
$16,000
$220,967.42$441,934.84$662,902.26$883,869.68$1,104,837.10$1,325,804.52$1,546,771.94$1,767,739.36$1,988,706.78$2,209,674.20
$0
$200
$400
$600
$800
$1,000
$1,200
$1,400
$1,600
$1,800
$2,000
Gross.Adds
Net Marketing
Contribution
Net Marketing Returns $MAnnual New Subscribers M
Total Marketing Spend $M
Current Spend
The chart below is a simulation of annual New Subscribers and net returns at various
total spending levels, and assuming the current mix of spending. As shown,
Company can expect to generate positive net marginal returns to marketing spending
up to an overall increase of +100%. By implication, not only is marketing spending
profitable, but will continue to be so at significantly higher spending levels.
21. • When optimizing marketing spend across media and local spend
across regions, we find considerable upside for Company to
improve marketing impact at current spending levels.
• Our solution calls for substantial spending increases for branded
message media, new product media and product promotion, while
cutbacks are called for all other media forms. Across the regions,
our solution also calls for local spending increases on local media
(product promotion and secondary media ) for the NE and Central
regions with cutbacks in the other two regions. Optimization has
identified an upside of +5.9% in New Subscribers at constant total
spending.
Marketing Spending Optimization
22. Optimized Local Marketing Spending
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Subscriber
Contribution
Current Spend Optimal Spend
West
Central
SE
NE
The company spends about $219MM of its total budget on local marketing efforts.
Despite our overall optimized solution for -20% less local spending, the Company can
actually achieve an additional +1.4 percent subscriber growth by allocating funds
more efficiently across the regions. This solution calls for increases in the NE and
Central regions, with relative reductions in the other two
23. Conclusions and Recommendations
The modeling exercise conducted for this telecom company has afforded us the
opportunity to learn considerably about the marketing dynamics of this company
and the telecom business.
We learned that, with very heavy spending, this business is highly marketing
dependent, and especially with respect to network TV advertising. Overall,
company marketing efforts drive 65.9 percent of total new subscribers over the
past year.
We learned that with the high spending and high frequency of commercials, that
media message management and optimizing media message is a critical part of
the marketing optimization equation. For this client, we determined that
branded messages were significantly more effective than non-branded products,
plans and services messages. As a consequence, the company should
increase its investment in branded messages, and doing so has the upside of
driving +4.3 percent higher new subscribers.
We learned that with a large number of media commercials, media execution
frequency and clutter poses a risk to advertising impact and effectiveness.
More non-branded media messages, in particular, pose a downside risk on
media effectiveness.
We learned that media quality is a measurable and important component of the
overall equation for media effectiveness.
24. Conclusions and Recommendations
The modeling also uncovered that customer satisfaction is important and is
measurable driver of business performance for the company,
We learned that the company’s overall marketing investment generates a
strong and positive ROI at 71 percent, but not all marketing channels
generate positive returns. While national network TV and product promotion
generate positive returns, the returns for print, OOH, radio, Spot TV and
online advertising fall short of generation positive returns.
Through our modeling efforts, we also determined that the company has a
two-fold opportunity for driving higher growth in subscribers through
optimizing its marketing spending. Across the national marketing mix, the
solution calls for higher spending in branded, non-branded and new-product
advertising plus product promotion. By contrast, this solution also calls for
a relative reduction in the less profitable media channels of print, OOH,
radio, Spot TV and online media. Such a solution is expected to generate
growth of +5.9 percent in new subscribers.
The second dimension and opportunity estimates that the company can
generate growth of +1.4 percent by optimizing its local media spending.
This calls for higher relative spending in the NE and Central regions and
relative reductions in the SE and West Region.
25. Model Validations
The following shows actual and model fit charts for each
of the four regions.
Overall, the fits and predictive capabilities of these
models were extremely high. Total model R-squares
averaged 98.8%. Holdout R-squares across 13 weeks
of data were 99.6%, showing a high degree of predictive
accuracy for these models. Total Mean Absolute Percent
Errors across all models is +/- 2.0%.