Forecasting New Product Sales Notes for Product Development

Forecasting New Product Sales Notes for Product Development Roger Bohn Material from various sources including mktgweb. wharton. upenn. edu/hutch/RWWI-HDTV. PPT See also: New Products Management by Crawford & di Benedetto Rev: Nov. 6 2013

Schedule of remaining classes n n n Today Nov. 6: Market penetration modeling. Reading posted on the web. Wed. Nov 13 Milestone: Design review. Concept testing and Design for Manufacturability (if time) Read: Chapters 9, chapter 13 Mon Nov. 18 Industrial design (Donald Norman Design of Everyday Things) Wed. Nov. 20 Milestone: Final prototype Mon Nov. 25 Studying old presentations; how to make an effective pitch. Wed. Nov. 27 or Sunday Dec. 1 Presentation rehearsals (90 minutes per team) Mon. Dec. 2 The Mythical Man Month (no preparation needed, but book is good) Tues. Night of Dec. 3: Milestone: Final presentations Wed. Dec 4: Identifying weak areas to improve for the business plan/final report Friday Dec 6 Milestone: Business plan due Wed. Dec 14 Milestone: Case studies due. 2

Personal journals 3

Key inputs each quarter n Sales volume (physical units) Q(t) n n n n Size of market Market share (t) Penetration curve (t) ASP at factory gate P Manufacturing costs = C(Q) Development costs, duration Discount rate 4

Sample spreadsheet for NPV 5

CUMULATIVE PENETRATION Typical Diffusion Curves: Cumulative Penetration SOURCE: Best (1997), Market-Based Management 6

Why Financial Analysis for New Products is Difficult n n n Target users don’t know. If they know they might not tell us. Poor execution of market research. Market dynamics. Uncertainties about marketing support. n n n Biased internal attitudes. Poor accounting. Rushing products to market. Basing forecasts on history. Technology revolutions.

Three types of sales models (at least) n n Steady state models: ultimate market share ATAR Dynamic models: adoption over time n n Multi-generation models n n “Diffusion of innovation” Social level: new type of product Curve-fitting to actual patterns How fast does new tech replace old equivalent? Ecosystem models? 8

1. Steady state: established non-durables (ATAR) n n n MS = T x R x Aware x Available Sales = Market size x MS = (# users) x (purchase rate per yr) MS = market share T = percent of “all buyers” who try product R = long run repeat rate 9

Figure 8. 5 A-T-A-R Model of Units Sold = Number of buying units x % Aware of product x % who would Try product if they can get it x % to whom product is Available x % of triers who become Repeat purchasers x Number of units each customer buys in a year Sales = Market size x Market Share = (# users) x (purchase rate per yr)

The A-T-A-R Model: Definitions n n n Figure 8. 6 Buying Unit: Purchase point (person or department/buying center). Aware: Has heard about the new product with some characteristic that differentiates it. Available: If the buyer wants to try the product, the effort to find it will be successful (expressed as a percentage). Trial: Usually means a purchase or consumption of the product. Repeat: The product is bought at least once more

Getting the Estimates for A-T-A-R Model xx: Best source for that item. x: Some knowledge gained. Figure 8. 7

Using field responses for trial rate

2. Diffusion models (population adoption of innovation) New Adopters = (i. e. , first purchases) [ p + q x (Old Adopters/Population) ] x [Population - Old Adopters] Based on models of contagious diseases, the Bass (1969) model of the diffusion of innovation has proved to be a simple & robust model of product life cycles. n likelihood of "spontaneous" innovation likelihood of "imitation" or "contagion" remaining market potential Average Diffusion Curves 100% 80% First Purchases Repeat = 20% Repeat = 50% Repeat = 100% 60% 40% 20% 0% Time Total users(T) = (new adopters(t) x repeat purchasers) 14

Parameters differ by market 15

Diffusion is faster when: Average Diffusion Curves 100% 80% 60% S-shaped Diffusion Curves 100% 40% 20% 0% 80% Time 60% 40% 20% 1. marketing expenditures are high, 2. competitive activity is high and products use similar technologies (why? ) 3. the innovation is simple, low risk, easy to observe, easy to try, 4. it is compatible with consumer knowledge and social values, 5. it has a clear relative advantage. 0% Time 16

Growth of your market share Average Diffusion Curves 100% New product in established market: 1. Diffusion curves still relevant, but top is your steady state share eg 12% 2. Parameters of curve depend on: n n number of competitors relative marketing campaigns; marketing strategy eg pull vs push 80% 60% 40% 20% 0% Time Of course, world never reaches steady state (equilibrium = a useful economic fiction) 17

Meta-Analysis of Diffusion Curves Meta-Analysis of 213 Product Innovations 100. 0% 90. 0% Market Penetration (i. e. , Cumulative Adoption) 80. 0% 70. 0% 60. 0% 50. 0% 40. 0% 30. 0% 20. 0% 10. 0% 1 3 5 7 9 Medical & Industrial innovation coefficient = imitation coefficient = rate = shape = 4. 6% 58. 5%. 16. 28 11 13 Years 15 Average 3. 9% 30. 2%. 11. 36 17 19 21 23 Consumer Durables 2. 3% 24. 0%. 07. 31 SOURCE: Sultan, Farley & Lehmann (1990), Journal of Marketing Research 18

3. Multi-Generational Diffusion Curves: Annual Sales of DRAM Memory Chips Six Generations of Dynamic RAM SOURCE: Best (1997), Market-Based Management 19

What drives diffusion/adoption of an innovation ? n Spontaneous Adoption Individuals adopt on the basis of their personal needs and marketing activities without regard to cumulative penetration (e. g. VOIP). -- classic marketing n Contagion or Imitation* Individuals become more likely to adopt the innovation as cumulative penetration increases because their observation of prior adopters and word-of-mouth information is important in their decision making (e. g. , personal digital assistants). -- “word of mouth”, market pull n Direct Network Externalities** Individuals become more likely to adopt the innovation as cumulative penetration increases because: the value of the innovation to each adopter increases as the number of adopters increases (e. g. , fax machines) -- niche markets; standards battles n Indirect Network Externalities** Individuals become more likely to adopt the innovation as cumulative penetration increases because the value of the innovation to the adopter depends on the availability of complementary products and the providers of these products respond to increasing demand (e. g. , HDTV; MM PCs; BB internet) -- market to software makers; standards battles Marketing approach differs for each 20

HOWEVER, diffusion curves are difficult to predict prior and just after launch n most products initially show increasing sales n marketing efforts and trial purchase may cause quick growth regardless of longrun diffusion n Few data points Concave (and Rapid) Diffusion Average Diffusion Curves 100% 80% S-shaped (and Slow) Diffusion Curves 100% 80% 60% 60% 40% 40% 20% 20% 0% Time First Purchases Repeat = 20% Repeat = 50% Repeat = 100% 0% Time 21

Direct and Indirect Network Externalities MANUFACTURERS (Supply Hardware) ed at ce i ed den M n t ke epe r Ma terd In COMPLEMENTORS (Supply Software, Content, & Ancillary Services) Opportunity Assessment, Forecasting & Planning Di r De ect D pe em nd a en nd ce Direct Demand Dependence CONSUMERS (Demand Whole Product) Professor Wes Hutchinson, The Wharton School, University of Pennsylvania

Ecosystems: Indirect Network Externalities Opportunity Assessment, Forecasting & Planning Professor Wes Hutchinson, The Wharton School, University of Pennsylvania

Is forecasting impossible? n For genuine revolutions, yes n n But 99% of “revolutionary” products are similar to predecessors Gather early data to pin down model n n Which model best fits? Similar historical products Model parameters Concept testing

End 25

“Standard paper template” Background: Industry, process description, key success factors (why learn? About what? ) Case A: Methods of learning May be historical, or present. Usually several complementary methods, 1 -2 of which are your focus Analyze: Strengths, weaknesses, etc. What it’s good for; not good for; attributes like cost Underlying changes: is target static, or dynamic? Case B: Revised method What it is; how it differs strengths/weaknesses by comparison Effects on rate of learning (claimed and actual) Analysis How to learn even faster (opportunities) How competitors/adversaries respond predictions 26