Advertising analysis
Lucy98765
W8.Periods2AdBrandAttitudes.pdf
Brand Management
Advertising and Brand Attitudes
P R O F. M A X J O O
Objectives Does advertising influence consumers’ attitudes toward a brand?
◦ Perceived quality ◦ Perceived value ◦ Recent satisfaction
How do we estimate the effects?
2
Brand attitude Consumers’ positive/negative association with a brand
◦ Is the brand associated with “good quality”? ◦ Is the brand associated with “good value”? ◦ Is the brand associated with “satisfaction”?
3
Brand attitude Consumers’ positive/negative association with a brand
◦ Is the brand associated with “good quality”? ◦ Is the brand associated with “good value”? ◦ Is the brand associated with “satisfaction”?
4
Brand attitude Consumers’ positive/negative association with a brand
◦ Is the brand associated with “good quality”? ◦ Is the brand associated with “good value”? ◦ Is the brand associated with “satisfaction”?
Uniform quality, credibility and experience beyond a single product ◦ Positive attitudes offer competitive advantages ◦ Readily available from GfK, Millward Brown, TNS and YouGov ◦ Managers track a brand’s health over time using brand attitude surveys
5
Why do we care? Brand attitudes
◦ Predict measurable lower-funnel metrics like sales and online searches ◦ Create differentiation and reduce pricing pressure ◦ Are inherently valuable, as reflected in valuations ◦ Might be a useful proxy to set ad budgets, especially for advertisers who
can’t estimate direct effects of ads on sales
6
Why do we care? Brand attitudes
◦ Predict measurable lower-funnel metrics like sales and online searches ◦ Create differentiation and reduce pricing pressure ◦ Are inherently valuable, as reflected in valuations ◦ Might be a useful proxy to set ad budgets, especially for advertisers who
can’t estimate direct effects of ads on sales
Advertising potentially influences brand attitudes, the brand attitudes may influence choice
◦ It offers an intermediate or surrogate measure of marketing effectiveness ◦ Facebook’s brand lift ◦ Scientific research has been skeptical
7
Du, Joo, Wilbur (2019) How do brand attitudes change with own and competitor ads?
How do these relationships vary across attitudes and ad media?
Can brand attitudes be attributed to ads?
8
Data 575 established brands/ 37 industries/ 252 weeks/ $264B ad spend
◦ Meta-analytic scope without publication bias ◦ 37% of national ad spend, over 10 million surveys
YouGov BrandIndex ◦ A panel of more than 1.5 million US consumers ◦ Each panelist completes up to one survey each month ◦ We focus on the following three questions:
◦ “Which of the brands do you associate with good quality?” ◦ “Which of the brands do you associate with good value-for-money?” ◦ “Would you identify yourself as a recent satisfied customer of any of these brands?”
9
Data 575 established brands/ 37 industries/ 252 weeks/ $264B ad spend
◦ Meta-analytic scope without publication bias ◦ 37% of national ad spend, over 10 million surveys
YouGov BrandIndex
10
Data 575 established brands/ 37 industries/ 252 weeks/ $264B ad spend
◦ Meta-analytic scope without publication bias ◦ 37% of national ad spend, over 10 million surveys
Kantar Media Stradegy ◦ Comprehensive ad placement and expenditure data ◦ #1 in competitive advertising intelligence
11
Data 575 established brands/ 37 industries/ 252 weeks/ $264B ad spend
◦ Meta-analytic scope without publication bias ◦ 37% of national ad spend, over 10 million surveys
Ford
12
0
10
20
30
40
50
60
Po si
tiv e
B ra
nd A
tti tu
de (%
)
3− M
ar −0
8 5−
M ay −0
8 7−
Ju l−
08 8−
Se p−
08 10 −N
ov −0
8 12 −J
an −0
9 16 −M
ar −0
9 18 −M
ay −0
9 20 −J
ul −0
9 21 −S
ep −0
9 23 −N
ov −0
9 25 −J
an −1
0 29 −M
ar −1
0 31 −M
ay −1
0 2−
A ug −1
0 4−
O ct −1
0 6−
D ec −1
0 7−
Fe b−
11 11 −A
pr −1
1 13 −J
un −1
1 15 −A
ug −1
1 17 −O
ct −1
1 19 −D
ec −1
1 20 −F
eb −1
2 23 −A
pr −1
2 25 −J
un −1
2 27 −A
ug −1
2 29 −O
ct −1
2
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ● ●
●
●
●
● ●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
● ●
●
●
● ●
●
● ●
●
●
● ● ● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
● ●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
0
10
20
30
40
50
60
A dv
er tis
in g
E xp
en di
tu re
($ M
ill io
n)
● Quality Value Satisfaction
Data 575 established brands/ 37 industries/ 252 weeks/ $264B ad spend
◦ Meta-analytic scope without publication bias ◦ 37% of national ad spend, over 10 million surveys
Toyota
13
0
20
40
60
Po si
tiv e
B ra
nd A
tti tu
de (%
)
3− M
ar −0
8 5−
M ay −0
8 7−
Ju l−
08 8−
Se p−
08 10 −N
ov −0
8 12 −J
an −0
9 16 −M
ar −0
9 18 −M
ay −0
9 20 −J
ul −0
9 21 −S
ep −0
9 23 −N
ov −0
9 25 −J
an −1
0 29 −M
ar −1
0 31 −M
ay −1
0 2−
A ug −1
0 4−
O ct −1
0 6−
D ec −1
0 7−
Fe b−
11 11 −A
pr −1
1 13 −J
un −1
1 15 −A
ug −1
1 17 −O
ct −1
1 19 −D
ec −1
1 20 −F
eb −1
2 23 −A
pr −1
2 25 −J
un −1
2 27 −A
ug −1
2 29 −O
ct −1
2
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
● ●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
● ●
●
● ●
● ●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
● ●
● ●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
0
10
20
30
40
50
A dv
er tis
in g
E xp
en di
tu re
($ M
ill io
n)
● Quality Value Satisfaction
Data 575 established brands/ 37 industries/ 252 weeks/ $264B ad spend
◦ Meta-analytic scope without publication bias ◦ 37% of national ad spend, over 10 million surveys
Coke
14
0
20
40
60
Po si
tiv e
B ra
nd A
tti tu
de (%
)
3− M
ar −0
8 5−
M ay −0
8 7−
Ju l−
08 8−
Se p−
08 10 −N
ov −0
8 12 −J
an −0
9 16 −M
ar −0
9 18 −M
ay −0
9 20 −J
ul −0
9 21 −S
ep −0
9 23 −N
ov −0
9 25 −J
an −1
0 29 −M
ar −1
0 31 −M
ay −1
0 2−
A ug −1
0 4−
O ct −1
0 6−
D ec −1
0 7−
Fe b−
11 11 −A
pr −1
1 13 −J
un −1
1 15 −A
ug −1
1 17 −O
ct −1
1 19 −D
ec −1
1 20 −F
eb −1
2 23 −A
pr −1
2 25 −J
un −1
2 27 −A
ug −1
2 29 −O
ct −1
2
●
● ●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
● ●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
● ●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
● ●
●
●
● ●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
0
10
20
30
A dv
er tis
in g
E xp
en di
tu re
($ M
ill io
n)
● Quality Value Satisfaction
Data 575 established brands/ 37 industries/ 252 weeks/ $264B ad spend
◦ Meta-analytic scope without publication bias ◦ 37% of national ad spend, over 10 million surveys
Apple
15
0
20
40
60
Po si
tiv e
B ra
nd A
tti tu
de (%
)
3− M
ar −0
8 5−
M ay −0
8 7−
Ju l−
08 8−
Se p−
08 10 −N
ov −0
8 12 −J
an −0
9 16 −M
ar −0
9 18 −M
ay −0
9 20 −J
ul −0
9 21 −S
ep −0
9 23 −N
ov −0
9 25 −J
an −1
0 29 −M
ar −1
0 31 −M
ay −1
0 2−
A ug −1
0 4−
O ct −1
0 6−
D ec −1
0 7−
Fe b−
11 11 −A
pr −1
1 13 −J
un −1
1 15 −A
ug −1
1 17 −O
ct −1
1 19 −D
ec −1
1 20 −F
eb −1
2 23 −A
pr −1
2 25 −J
un −1
2 27 −A
ug −1
2 29 −O
ct −1
2
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
● ●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
● ●
●
●
● ●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
0
10
20
30
A dv
er tis
in g
E xp
en di
tu re
($ M
ill io
n)
● Quality Value Satisfaction
Regression finds Own national traditional ad spend increases all three brand attitude metrics
Own local traditional ad spend tends to improve perceived quality and perceived value, but not recent satisfaction
Own digital ad spend increases perceived value, but not other two
Your ad spend influences consumers’ attitudes toward your brand ◦ This updates the existing knowledge that ads cannot
16
Regression finds Competitors’ national traditional ad spend negatively impacts perceived quality, but not perceived value or recent satisfaction
Competitors’ local traditional ad spend tends to reduce all three brand attitude metrics
Competitors’ digital ads tend to reduce perceived quality and recent satisfaction
If your competitors spend on ads but you do not, consumers’ attitudes toward your brand are negatively influenced
17
Takeaways Advertising helps to formulate positive attitudes toward a brand beyond awareness
Competitive advertising may hurt
These effects are observed in various types of media
18
Now let’s do it Make sure to be ready to use Radiant using one of the three options:
1. Install on your computer: https://radiant-rstats.github.io/docs/install.html 2. Log on to the virtual lab at http://ucr.apporto.com/ using your UCR Net ID
and remotely work on Radiant from the virtual lab ◦ Instructions for the virtual lab is here: https://ucrsupport.service-
now.com/ucr_portal/?id=kb_article&sys_id=8b5964291b84d49026bd635bbc4bcbd7 ◦ In case that you have trouble working on the virtual lab, you will need to directly contact the IT
office
3. (Emergency protocol, but not recommended) Use online version of Radiant at https://vnijs.shinyapps.io/radiant/
◦ Functionality is limited, and security is a concern
Download your ad-brand dataset from eLearn, Course Materials
19
Radiant First run “Rstudio”
20
Radiant Click “Addins”
21
Radiant Click “Start radiant (browser)”, then Radiant will show up on your web browser
22
Radiant On the left panel, click the drop down menu under “Load data of type”, then choose “csv”
23
Radiant Click “Load”, then select the folder you stored the data file.
Select the file (ad_brand_data_reg.csv) then click “Select”
24
Your data Ad_brand_data_reg.csv
◦ Brand: brand identifier (anonymized) ◦ Industry: industry that a brand belongs to (anonymized) ◦ time_period: exact time that the data point was measured ◦ Yrqtr: Year and quarter of “time_period” ◦ qua: % of respondents who associated the brand with “good quality” ◦ val: % of respondents who associated the brand with “good value” ◦ sat: % of respondents who associated the brand with “satisfaction” ◦ nat_1m: national media advertising in million dollars ◦ loc_1m: local media advertising in million dollars ◦ dig_1m: online advertising in million dollars ◦ comp_...: competitor advertising in the same industry
25
Regression analysis Click “Model” on top menu, and select “Linear regression”
26
Regression analysis: Step 1 Select “qua” as “Response variable”
Select “nat_1m”, “loc_1m”, and “dig_1m” as “Explanatory variables”
Click “Estimate model”
Coefficient for “nat_1m” is 0.004, with “***” ◦ What does it mean? ◦ Unit increase in national ads ($million) is associated with 0.004 unit increase
in % of people who perceive the focal brand as good quality ◦ “***” on the right hand side indicates that
◦ Coefficient is statistically distinguishable from zero
27
Regression analysis: Step 1 Select “qua” as “Response variable”
Select “nat_1m”, “loc_1m”, and “dig_1m” as “Explanatory variables”
Click “Estimate model”
Coefficient for “loc_1m” is -0.000, without “*” ◦ What does it mean? ◦ Unit increase in local ads ($million) does not exhibit a relationship with % of
people who perceive the focal brand as good quality ◦ No “*” on the right hand side indicates that
◦ Coefficient is NOT statistically distinguishable from zero
28
Regression analysis: Step 1 Select “qua” as “Response variable”
Select “nat_1m”, “loc_1m”, and “dig_1m” as “Explanatory variables”
Click “Estimate model”
Coefficient for “dig_1m” is 0.011, with “*” ◦ What does it mean? ◦ Unit increase in online ads ($million) is associated with 0.011 unit increase in
% of people who perceive the focal brand as good quality ◦ “*” on the right hand side indicates that
◦ Coefficient is statistically distinguishable from zero
29
Regression analysis: Step 2 Select “qua” as “Response variable”
Select “nat_1m”, “loc_1m”, and “dig_1m” as “Explanatory variables”
Add “Industry” and “yrqtr” to the “Explanatory variables” ◦ Ctrl + Click multiple variables
Click “Re-estimate model”
How did coefficients change? ◦ All three coefficients are statistically distinguishable from zero with control
variables! ◦ Now how can we interpret the coefficients?
30
Interpretation, on average $1m increase in national ads is associated with 0.3% (=0.003) increase in the % of people who perceive the focal brand as good quality
$1m increase in local ads is associated with 0.2% (=0.002) increase in the % of people who perceive the focal brand as good quality
$1m increase in digital ads is associated with 1.1% (=0.011) increase in the % of people who perceive the focal brand as good quality
Control variables help researchers obtain more accurate coefficients
31
Assignment 8: Due May 23rd Carefully review the slides in pages 18 ~ 30, and run the following regression analyses
1. Perceived quality ◦ Regress qua on nat_1m, loc_1m, and dig_1m, and interpret coefficients ◦ Regress qua on nat_1m, loc_1m, dig_1m, industry, and yrqtr, and interpret coefficients ◦ Discuss how control variables (industry and yrqtr) influenced advertising effectiveness coefficients
2. Perceived value ◦ Regress val on nat_1m, loc_1m, and dig_1m, and interpret coefficients ◦ Regress val on nat_1m, loc_1m, dig_1m, industry, and yrqtr, and interpret coefficients ◦ Discuss how control variables (industry and yrqtr) influenced advertising effectiveness coefficients
3. Satisfaction ◦ Regress sat on nat_1m, loc_1m, and dig_1m, and interpret coefficients ◦ Regress sat on nat_1m, loc_1m, dig_1m, industry, and yrqtr, and interpret coefficients ◦ Discuss how control variables (industry and yrqtr) influenced advertising effectiveness coefficients
4. Repeat regressions in 1~3 with competitive advertising (compnat_1m, comploc_1m, and compdig_1m) and discuss how control variables help
Note: Please attach screenshot of each work
32
