1. Forecasting: Using A Meta-Marketing Machine Ted Mitchell

2. Improve Forecasting By Using Two Performances rather than one Performance 1) We have tried the average of the two or more performances 2) Now Use the differences between the two performances as the measure of Input and the measure of output and the meta-conversion rate in the Meta-Marketing Machine

3. Use the Differences in the Input, ∆I, and the Outputs, ∆O to construct a Meta-Marketing machine Performance P1Performance P2 Meta-Marketing Machine Input: Advertising Dollars, A A1 = $2,000A2 = $2,800 Calculate the conversion rate Pies per dollar of advertising, r = Q/A Q/A= r1= 12,000/$2,000 = 6 pies per dollar of advertising conversion rate, r2 = 13,200/$2,800 = 4.7 pies per dollar of advertising Output: Sales of Pies, Q Q1 = 12,000 piesActual Sales Q2 = 13,200 pies

4. Have Constructed a Meta-Advertising Machine Meta-Advertising Machine Input: Advertising Dollars, A ∆A = $800 Calculate the conversion rate m = ∆Q/∆A m = 1,200/$800 m = 1.5 more pies for every extra dollar spent Output: Sales of Pies, Q ∆Q = 1,200 pies

5. Set up the Meta-Advertising Machine for making a forecast Meta-Advertising Machine Input: Advertising Dollars, A ∆A = $800 Calculate the conversion rate Pies per dollar of advertising, r = Q/A m = ∆Q/∆A m = 1,200/$800 m = 1.5 more pies for every extra dollar spent Output: Sales of Pies, Q ∆Q = 1,200 pies

6. The Direct Relationship Between Differences A Direct relationship between two variables, Y and X, is defined as one where the ratio of the two variables has a constant value, k Y/X = k A Variable can be also be defined as the difference in the amount of a variable, ∆Y = Y 2 – Y 1 ∆X = X 2 – X 1 A Direct relationship between the differences of two variables, ∆Y and ∆X is defined as ∆Y/∆X = m The constant, m, is defined as a meta-conversion rate. Slope Origin form of the direct relationships is ∆Y = m x ∆X

7. Input and Output of a Meta-Marketing Machine Input Factor, ∆A Output, ∆Q 0, 0 ∆A = $800 ∆Q = 1,200 pies X Calibration point (∆A, ∆Q) = ($800, 1,200 pies) Rise Run

8. Input and Output of a Meta-Marketing Machine Input Factor, ∆A Output, ∆Q 0, 0 ∆A = $800 ∆Q = 1,200 pies X Calibration point (∆A, ∆Q) = ($800, 1,200 pies) Rise Run Meta-Conversion rate m = ∆Q/∆A M = 1.5 pies per dollar increase Meta-Conversion rate m = ∆Q/∆A M = 1.5 pies per dollar increase

9. Calculate the Meta-conversion rate, m = ∆O/∆I Performance 1Actual Performance 2 ∆P = P2-P1 Input: Advertising Dollars, A A1 = $2,000A2 = $2,800∆A = $800 Calculate the conversion rate Pies per dollar of advertising, r = Q/A Ignore this Q/A= r1= 12,000/$2,000 = 6 pies per dollar of advertising Ignore this conversion rate, r2 = 13,200/$2,800 = 4.7 pies per dollar of advertising Meta-conversion rate, m = ∆Q/∆A = 1,200/$800 = 1.5 change in pies for a $1 change in advertising Output: Sales of Pies, Q Q1 = 12,000 piesActual Sales Q2 = 13,200 pies ∆Q = 1,200 pies

10. Slope-Origin Equation of Two- Factor Meta-Marketing Machine Input Factor, ∆A Output, ∆Q 0, 0 ∆A = $800 ∆Q = 1,200 pies X Calibration point (∆A, ∆Q) = ($800, 1,200 pies) Rise Run Meta-Conversion rate or slope m = rise/run m = ∆Q/∆A Meta-Conversion rate or slope m = rise/run m = ∆Q/∆A Slope-Origin Equation for Two-Factor Meta-Machine ∆Q = m x ∆A Slope-Origin Equation for Two-Factor Meta-Machine ∆Q = m x ∆A

11. Forecast Example Your boss wants to increase the current advertising budget from $800 to $1,000 A proposed change of ∆A = $200 in advertising, You know the meta-conversion rate has been calibrated as an increase of m = 1.5 pies for every dollar spent on advertising What is the forecasted number of pies sold when the proposed change in advertising effort occurs

12. Set up the Meta-Advertising Machine for making a forecast Meta-Advertising Machine Proposed change in the size of the advertising Forecasted Output Input: Advertising Dollars, A ∆A = $800∆A = $200 Calculate the conversion rate Pies per dollar of advertising, r = Q/A m = ∆Q/∆A m = 1,200/$800 m = 1.5 more pies for every extra dollar spent Use the calibrated meta- conversion rate m = 1.5 more pies per dollar spent m = 1.5 more pies per dollar spent Output: Sales of Pies, Q ∆Q = 1,200 pies∆Q = m x ∆A ∆Q = 1.5 x $200 ∆Q = 300 pie increase

13. #1 Sample Forecasting Question Market research has calibrated the meta-conversion rate, m, for the relationship between the difference in the number of pies sold, and the difference in the amount of advertising being used as Meta-conversion rate, m = 1.5 pies for a change of $1 of advertising expense Your boss has proposed an increase of $200 in advertising expense. What is the anticipated change in the number of pies that will be sold? Answer ∆Q = m x ∆π ∆Q = 1.5 pies per dollar x $200 ∆Q = a change of 300 pies

14. Forecasting with the Meta-Marketing Machine Calibrated on Two Observed Performances Meta-Marketing Machine Forecasted Performance Input: Advertising Dollars, ∆A ∆AProposed change in advertising ∆A=$200 Calibrated meta- conversion rate Pies per dollar of advertising, m = ∆Q/∆A m = ∆Q/∆A = 1,200/$800 = 1.5 change in pies for a $1 change in advertising Using the calculated meta- conversion rate ∆Q/∆I = 1.5 Output: Sales of Pies, Q ∆QForecasted change in Sales ∆Q = 1.5 x $200 ∆Q = a change of 300 pies sold

15. #2) Second Version Forecasting Question Market research has calibrated the meta-conversion rate, m, for the relationship between the difference in the number of pies sold, and the difference in the amount of advertising being used as Meta-conversion rate, m = 1.5 pies for a change of $1 of advertising expense You are currently selling Q1 = 800 pies with an advertising expenditure of A1 = $1,000. Your boss wants to increase the current advertising budget to a new total of A2 = $1,200. What is the forecasted number of total pies sold, Q2? Answer ∆Q = m x ∆A ∆Q = 1.5 pies per dollar x ∆A (Q2 – Q1)= 1.5 pies per dollar x (A2 – A1) (Q2 – 800) = 1.5pies per dollar ($1,200 – $800) Q2 = 800 + 1.5(400) Q2 = 800 + 600 = 1,400 pies sold ∆Q = Q2 – Q1 = a change of 600 pies

16. Forecasting with the Meta-Marketing Machine Calibrated on Two Observed Performances Meta-Marketing MachineCurrent Performan ce Proposed Level Forecasted Change Performance Forecasted new level of output Input: Advertising Dollars, ∆A A1 = $800 Proposed new level A2 =$1,200 Proposed change in advertising ∆A = A2 –A1 ∆A = $400 A2 = $1,200 Calibrated meta- conversion rate m = ∆Q/∆A m = ∆Q/∆A = 1,200/$800 m = 1.5 change in pies for a $1 change in advertising Using the calculated meta- conversion rate ∆Q/∆I = 1.5 (Q2-Q1)/(A2-A1) =1.5 Output: Sales of Pies, Q Q1 = 800 pies Forecasted new level, Q2 Forecasted change in Sales ∆Q = 1.5 x 400 ∆Q = 600 change in the quantity of pies sold Q2 = ∆Q + Q1 Q2 = 600 + 800 Q2 = 1,400 pies

17. Any Questions About Our review of Constructing a Meta-marketing Machine Forecasting with a slope-origin version of a meta-marketing machine Why is more accurate to use than a slope-origin equation calibrated from a single observation