Spreading versus focusing:

(1)

Spreading versus

focusing:

.

An analysis on the effectiveness of allocating

the marketing budget across one or multiple

advertising channels on customer mindset

metrics

(2)

The media landscape has changed…

What channels

should marketers

focus on?

What is the best

strategy?

(3)

(4)

The dataset (1)

91-week period

(2019-2020)

Six brands

Advertising

expenditures across

six channels

(5)

The dataset (2)

Television

Radio

Out of home

(6)

Gini

coefficient

Herfindahl

Index

Rosenbluth

index

Media concentration (1)

“Media concentration has been defined as the manner

at which the marketing budget of a brand is divided

(7)

Media concentration (2)

Spread equally across

(8)

Media concentration (2)

Spread equally across

all channels

Focus on one channel

(9)

Brand Advertising frequency (% of weeks with advertising) Average advertising investment (in € per week when investing) Average number of channels used during weeks with advertising* Advertising frequency using 1 channel (% of advertising weeks) Advertising frequency using > 1 channel (% of advertising weeks) Brand A 18.9% €85.126 1.12 (0.332) 88.2% 11.8% Brand B 46.7% €208.102 1.79 (1.000) 50.0% 50.0% Brand C 76.7% €194.075 1.61 (0.911) 59.4% 40.6% Brand D 23.3% €178.776 1.10 (0.301) 90.5% 9.5% Brand E 92.2% €140.631 1.10 (0.297) 90.4% 9.6% Brand F 55.6% €172.491 1.02 (0.141) 98.0% 2.0%

Model-free insights (1)

(10)

Model-free insights (1)

Predominantly

focus strategy

(11)

Model-free insights (1)

Indication

of spreading

(12)

Brand Television Magazines Out of home Newspapers Radio Brand A 52.90% 41.20% 11.80% 5.88% -Brand B 59.50% 21.40% 21.40% 7.14% 38.10% Brand C 95.70% 13.00% 20.30% 21.70% 10.10% Brand D 95.20% 9.52% 4.76% - -Brand E 98.80% 10.80% - - -Brand F 94.00% 8.00% - -

-Model-free insights (2)

(13)

-Model-free insights (2)

Television most

preferred

(14)

-Model-free insights (2)

Most diversity

in channel

usage

(15)

-Model-free insights (3)

(16)

-Model-free insights (3)

Television is

dominating the

advertising

expenditures

(17)

-Model-free insights (3)

Relatively

speaking, these

are spreaders

(18)

Model-free insights (3)

Brand A Brand B Brand C

Min Max Mean Sd Min Max Mean Sd Min Max Mean Sd RB 0.988 1 0.999 0.0030 0.415 1 0.918 0.145 0.456 1 0.905 0.160

HHI 0.998 1 0.999 0.0030 0.400 1 0.914 0.152 0.416 1 0.901 0.167

GINI 0.831 0.833 0.833 0.0005 0.598 0.833 0.811 0.0470 0.635 0.833 0.807 0.0491

Brand D Brand E Brand F

HHI 0.965 1 0.998 0.0077 0.858 1 0.998 0.0160 0.970 1 0.999 0.0043

(19)

Model-free insights (3)

HHI 0.998 1 0.999 0.0030 0.400 1 0.914 0.152 0.416 1 0.901 0.167

GINI 0.831 0.833 0.833 0.0005 0.598 0.833 0.811 0.0470 0.635 0.833 0.807 0.0491

Min Max Mean Sd Min Max Mean Sd Min Max Mean Sd RB 0.965 1 0.998 0.0075 0.867 1 0.998 0.0150 0.970 1 0.999 0.0042 HHI 0.965 1 0.998 0.0077 0.858 1 0.998 0.0160 0.970 1 0.999 0.0043 GINI 0.827 0.833 0.833 0.0013 0.808 0.833 0.833 0.0029 0.828 0.833 0.833 0.0007

Most variation in

the concentration

metrics

(20)

Model-free insights (3)

HHI 0.998 1 0.999 0.0030 0.400 1 0.914 0.152 0.416 1 0.901 0.167

GINI 0.831 0.833 0.833 0.0005 0.598 0.833 0.811 0.0470 0.635 0.833 0.807 0.0491

HHI 0.965 1 0.998 0.0077 0.858 1 0.998 0.0160 0.970 1 0.999 0.0043

(21)

Model choice

• Error Correction Model (ECM)

• Distinguish short-term and long-term effects

• Unit root test

• All variables should be stationary

• Levin, Lin and Chu’s (2002) panel unit-root test and the Im-Pesaran-Shin test (Im,

Pesaran & Shin, 2003) were highly significant

• Chow test

• Pooling is now allowed, thus unit-by-unit models

(22)

(23)

Aided brand

awareness

Full dataset

With

interaction

Without

Consideration

Full dataset

With

interaction

Without

(24)

Aided brand

awareness

Full dataset

Subset Brand B

_{and Brand C}

With

interaction

Without

interaction

With

interaction

Without

Consideration

Full dataset

Subset Brand B

_{and Brand C}

With

interaction

Without

interaction

With

interaction

Without

(25)

Aided brand

awareness

Full dataset

Subset Brand B

_{and Brand C}

With

interaction

Without

interaction

With

interaction

Without

Consideration

Full dataset

Subset Brand B

_{and Brand C}

With

interaction

Without

interaction

With

interaction

Without

(26)

(27)

(28)

Aided brand

awareness

Full dataset

Subset Brand B

_{and Brand C}

With

interaction

Without

interaction

With

interaction

Without

Consideration

Full dataset

Subset Brand B

_{and Brand C}

With

interaction

Without

interaction

With

interaction

Without

(29)

Results (5.1.3)

“…the significant long-term

effects that were present in the

(30)

Aided brand

awareness

Full dataset

Subset Brand B

_{and Brand C}

With

interaction

Without

interaction

With

interaction

Without

Consideration

Full dataset

Subset Brand B

_{and Brand C}

With

interaction

Without

interaction

With

interaction

Without

(31)

(32)

(33)

Aided brand

awareness

Full dataset

Subset Brand B

_{and Brand C}

With

interaction

Without

interaction

With

interaction

Without

Consideration

Full dataset

Subset Brand B

_{and Brand C}

With

interaction

Without

interaction

With

interaction

Without

(34)

Results (5.2.3)

“…the significant effects of media

concentration and zero advertising

weeks have disappeared (compared

(35)

Additional analyses (1)

Distribution of activity is not

always captured by media

(36)

Aided brand

awareness

With

interaction

Without

Consideration

With

interaction

Without

Additional analyses (2)

“It might be interesting to repeat the analysis for

Brand B and Brand C by using the number of

(37)

Results (5.3.1)

“These findings suggest that there

are no effects of advertising and

the number of channels on aided

(38)

Results (5.3.2)

“…earlier models with a significant

no advertising dummy, showed a

(39)

Discussion

Advertising

expenditures

Brand awareness

Brand consideration

• No significant effects of advertising expenditures have been found

• Doubtful based on existing literature

(40)

Discussion

Advertising

expenditures

Brand awareness

Brand consideration

• One model for Brand B and Brand C suggests that zero advertising

weeks have a negative long-term effect

• Two models across all brands suggest that zero advertising weeks

have a positive long-term effect

(41)

Discussion

Advertising

expenditures

Brand awareness

Brand consideration

Media

concentration

• Only a significant and positive main effect in some models

• This suggests that focusing is a more effective strategy than spreading

• These effects did not hold in the subset for Brand B and Brand C and

were not present in the additional analyses (with number of

channels)

(42)

Managerial implications

• Opportunity to explore new channels

• Model-free insights suggest that IMC has not been widely adopted by

marketers

• For some brands, half of the channels remain unused

• A few findings suggest that focusing is a more effective strategy than

spreading

(43)

Limitations and future research directions

• Number of brands included and covered timespan are rather small.

Moreover, not much variation in media budget allocation.

• Only traditional media channels were included (e.g. digital could be

added)

• Only CPG brands were included

• Perhaps more variation in other categories

• Extension possible with other marketing mix tools

(44)