1. New Product Development
Sales Forecasting &
Financial Analysis
Moving down.
Recall that last week, we talked about internal idea screening.
This week we’re going to continue screening, but now with consumer input.
Should finish early so that groups can meet for the case

2. Sales Forecasts With Sales Potential Estimates LOGIT:
The customer always buys the product with the largest value of “true” preference
but, we have to measure this true preference and we do that imperfectly.
If we know something about the distribution of the error term, we can begin to
predict the probability that the “true” preference for the product is larger than the
true preference for other products. This prediction becomes some mathmatical
function based on the measured preferences.
Best used for incremental innovations

3. Mail Concept Test – Drawing / Diagram

4. Sales Potential Estimation
Translating Intent into Sales Potential
Example: Aerosol Hand Cleaner After examining norms for comparable existing products, you determine that:
90% of the “definites”
40% of the “probables”
10% of the “mights”
0% of the “probably nots” and “definitely nots” will actually purchase the product
Apply those %age to Concept Test results:
The “norms” depend on the type of product; how new it is; and whether any
incentive will be used at launch.
Goal: Translate

5. Sales Potential Estimation
Translating Intent into Sales Potential
Apply those %age to Concept Test results:
90% of the “definites” (5% of sample) = .045
40% of the “probables” (36%) = .144
10% of the “mights” (33%) = .033
0% of the last 2 categories = .000
Sum them to determine the %age who would actually buy: = .22
Thus, 22% of sample population would buy (remember: this % is conditioned on awareness & availability)
The “norms” depend on the type of product; how new it is; and whether any
incentive will be used at launch.
Goal: Translate

6. From Potential to Forecast
With Sales Potential Estimates:
To remove the conditions of awareness and availability, multiply by the appropriate percentages:
If 60% of the sample will be aware (via advertising, etc.) and the product will be available in 80% of the outlets, then:
(.22) X (.60) X (.80) = .11
11% of the sample is likely to buy

7. Sales Forecasts With Sales Potential Estimates
Diffusion of Innovations
The Bass Model:
Predicts pattern of trial (doesn’t include repeat purchases) at the category level
Works for all types of products, and can be used with discontinuous innovations
LOGIT:
The customer always buys the product with the largest value of “true” preference
but, we have to measure this true preference and we do that imperfectly.
If we know something about the distribution of the error term, we can begin to
predict the probability that the “true” preference for the product is larger than the
true preference for other products. This prediction becomes some mathmatical
function based on the measured preferences.
Best used for incremental innovations

8. The Bass Model
Estimates s(t) = sales of the product class at some future time t: s(t) = pm + [q-p] Y(t) - (q/m) [Y(t)]2 Where p = the “coefficient of innovation” [Average value=.04] q = the “coefficient of imitation” [Average value =.30] m= the total number of potential buyers Y(t) = the total number of purchases by time t

9. The Bass Model Important Feature
Once p and q have been estimated, you can determine the time required to hit peak sales (t*)
and the peak sales level at that time (s*): t* = (1/(p+q)) ln (q/p) s* = (m)(p+q)2/4q

10. Financial Analysis
Now, it’s time to use the forecasts to do financial projections:

11. Financial Analysis How Sophisticated? Early Stages:
Depends on the quality/reliability of the data and the stage you’re in
Early Stages:
Simple cost/benefit analysis or
“Sanity Check” as 3M uses:
attractiveness index = (sales X margin X (life).5 ) / cost
sales= likely sales for “typical year” once launched margin = likely margin (in percentage terms) life = expected life of the product in years (sq root discounts future) cost = cost of getting to market (dev., launch, cap.ex.)
Talk about warning of diving in too fast on financial analysis.
Many good projects get killed if applied too soon. Fuzzy is supposed to be fuzzy!
3M:
Computes attractiveness index and uses 2 arbitrary hurdle points: X and Y
If index exceeds X the project passes;
if index is between X and Y, more investigation is required
if index is below Y, it’s killed

12. Financial Analysis: Later Stages
Payback and Break-Even Times
Cycle Time
Payback Period
Break-Even Time (BET) = Cycle Time + Payback Pd.
Benefits:
1) Methods are simple and easy to understand
2) They capture both risk and return. A faster payback or BET means botha
higher return on investment and a lower risk
3) They use a cash flow approach, avoiding accounting method disputes
4) Projections only as far into the future as the point of cash recovery are needed
--- don’t need 10-yrs like you do for DCF and NPV

13. Financial Analysis: Later Stages
Payback and Break-Even Times
Benefits:
1) Methods are simple and easy to understand
2) They capture both risk and return. A faster payback or BET means botha
higher return on investment and a lower risk
3) They use a cash flow approach, avoiding accounting method disputes
4) Projections only as far into the future as the point of cash recovery are needed
--- don’t need 10-yrs like you do for DCF and NPV

14. Financial Analysis: Later Stages
Payback and Break-Even Times
Discounted Cash Flows (DCF, NPV, or IRR)
The most rigorous analysis for new products:
year-by-year cash flow projections discounted to the present
the discounted cash flows are summed
if the sum of the dcf’s > initial outlays, the project passes
The “Dark Side” of NPV (for NPD)
Unfairly penalizes certain projects by ignoring the Go/Kill options along the way (option values not accounted for in traditional NPV)
Benefits:
1)Recognizes that time is money -- penalizes projects where earnings are realized
late (late launch dates and revenue streams)
2) Also a cash flow method, avoiding accounting issues
3) Puts less emphasis on the cash flows that are the most difficult to estimate
THE DARK SIDE:
p.151 Cooper
Management can “Kill” projects along the way, but NPV theory treats it as if
the decision is “all or nothing”
*** When a project is a high-risk one -- when the prob. Of technical or
commercial success is low (define those success terms) OR when the costs
to undertake the project are high, NPV considerably understates the true
value of the project. You end up killing otherwise valuable projects ***
So, what are your other options…….

15. Financial Analysis: Later Stages
Payback and Break-Even Times
Discounted Cash Flows (DCF, NPV, or IRR)
Options Pricing Theory (OPT)
Recognizes that management can kill a project after an incremental investment is made
At each phase of the NPD process, management is effectively “buying an option” on the project
These options cost considerably less than the full cost of the project -- so they are effective in reducing risk
Kodak uses a decision tree and uses OPT to compute the Expected Commercial Value (ECV) of a given project
Used by Kodak -- they don’t use NPV

16. Using OPT to find the ECV
Commercial
Success
Pcs
$PVI
Technical Success
Yes
Launch $C
Pts
Yes No
Development $D
Commercial
Failure
$ECV No
Technical Failure
KEY:
Pts = Prob of tech success $D = Development costs remaining
Pcs= Prob of comm success $C = Commercialization/launch costs
$ECV = Expected commercial value $PVI = Present value of future earnings

17. Using OPT to find the ECV
ECV = [ [(PVI * Pcs) - C] * Pts] - D
KEY:
Pts = Prob of tech success $D = Development costs remaining
Pcs= Prob of comm success $C = Commercialization/launch costs
$ECV = Expected commercial value $PVI = Present value of future earnings

18. NPV vs. OPT: An Example TRADITIONAL NPV (no probabilities):
Income stream, PVI (present valued) $40 million
Commercialization costs (launch & captial) $ 5 million
Development costs $ 5 million
Probability of commercial success %
Probability of technical success %
Overall probability of success %
TRADITIONAL NPV (no probabilities):
= Decision = Go
NPV with probabilities:
(.25 X 30) - (.75 X 10) = Decision = Kill
ECV or OPT:
{ [(40 x .5) - 5] * .5} - 5 = Decision = Go