Optimization of TV SPOT Expenditure Distribution by Area

Case

We are sometimes requested to optimize ad expenditure distribution. Optimal distribution by multiple media, optimal placement timing within a year, and optimal spot expenditure by area are exemplary. Here, let’s suppose we have a client aiming for an increase in their product awareness nation wide using the process of optimizing spot expenditure distribution by area under the following conditions:

• Budget: 100 M yen
• Also purchased a national net of TIME

What is optimization?

In this case, optimization is defined as maximizing the awareness rate per unit of ad expenditure in the sum of all areas. In order to achieve this goal, there are two things to consider:

• Optimizing buying effectiveness: How to effectively purchase 1 GRP (Measured by cost/GRP, or in other words per cost)
• Optimizing Ad effectiveness: How to maximize the awareness rate of 1 GRP (Measured by awareness/GRP)
  By building a model of efficiency and effectiveness mathematically, we can identify the distribution maximum of awareness per unit in the total area.

Optimization Process

Optimization can be conducted in the following way:

1. Building a model of SPOT expenditure devotion to awareness rate by area by,

a. Building an "Expenditure -> GRP" model
b. Building a "GRP -> awareness" model
c. Building an "Expenditure -> awareness" model
2. Execute optimization using the above built models.

The following are instructions for 1 and 2.

1. Building a model of "SPOT Expenditure to awareness"
Here, we build 2 models, "Expenditure -> awareness" and "Expenditure -> GRP", and "GRP -> awareness", and mix the two at the very end for two reasons:

To reflect the difference of per cost of TIME and SPOT.
To show awareness from ad volume, as it is necessary to measure by GRP, not the expenditure
(a) Building a "Expenditure -> GRP" model
The price of 1 GRP varies with TIME, SPOT, and area. Therefore, the situations below could happen. (In TIME, the concept of per cost is not usually used, but for the comparison to SPOT, TIME per cost is calculated by the results from the past.)

Area A	Area B
SPOT per cost < TIME per cost -> Total per cost decreases as more and more SPOT is devoted	SPOT per cost > TIME per cost -> Total per cost increases as more and more SPOT is devoted

 

 

 

 

 

 

 

<Enlarge Image>

In this case, it can be concluded that devoting a lot of SPOT into area B is unwise in terms of cost efficiency.

(b) Building a "GRP -> awareness" model
Awareness increased by area varies even with 1 GRP. We can build a single regression analysis from the awareness data and GRP data from the past. Generally, the efficiency of awareness decreases as GRP devotion increases and therefore a non-linear (degression) model is appropriate. (The green line indicates efficiency in each GRP volume)

(c) Building a “Expenditure -> awareness Emodel
Next we can establish “the flow of Expenditure to awareness E combining the first two models by area. (TIME must be placed, therefore the model is under the assumption that the SPOT is devoted on top of that.)

When considering “buying efficiency Eand “ad effectiveness E the above result indicates that 1.2 M yen of SPOT Expenditure is the most efficient way of obtaining awareness. (However, this GRP may not fit when optimization includes other areas.)

2. Executing optimization using the built models.
By using model (c) for all areas, we can then optimize the distribution of the 100 M yen budget to each area. Here, optimization is executed under the condition that the awareness/cost (cost efficiency of awareness) of all areas total is maximized. The following table is the sample result (numbers are artificial):

 

In the above, the budget was fixed at 100 M yen. However, once the expenditure is changed by repeating the optimization process, the optimal devotion expenditure can be calculated. Also, optimization under certain conditions such as “Devoting a minimum of 500 GRP in every area Eor “Obtaining 50% awareness in the Kanto area Eis possible.

Written by Kenji Aoyagi, Senior Analyst

Please contact us with questions or for more detailed information.
spiindex@spi-consultants.net