New Product Forecasting Models Review
0 likes424 views
The document discusses several models for forecasting new product sales, including: - Parfitt & Collins model which predicts share based on trial, repeat rates, and buying rate. - Fourt-Woodlock model which estimates volume as a function of cumulative trial, repeat percentage rates by depth of repeat, and buying rate. - Assessor model which evaluates new product appeal among different consumer segments through in-person interviews. It also describes the Hendry market structure model which integrates factors like category share/loyalty, available switchers, and new product effectiveness to forecast share. Bases model measures purchase interest, transaction size, and frequency to estimate unit sales.
New Product Forecasting Models Review
- 1. New Product Forecasting Model’s Technical Review Art Christiani President The Hendry Partnership artchristiani@gmail.com 1-203-993-4823
- 2. 2 How to increase confidence in forecasting new products. 1. Key concepts 2. Forecasting Techniques Pre-Test Market Models • Hendry • Bases • Assessor • Regression Models – Top 100 / Attraction Models In-Market Models • The Parfitt & Collins Model • Fourt-Woodlock • TUP Models • Hendry 3. Measuring Cannibalization a. In-market measurement using store-level data b. Source of Volume analysis using panel data Forecasting – New Products
- 3. Concepts used in forecasting new products all involve consumer behavior, from awareness to purchase • Awareness - Consumers who are aware of the new product • Persuasion - Consumers who are persuaded to buy the product • Trial - Consumers who try the new product for the first time • ATC - Awareness to trial conversion • Repeat - Consumers who try the product for 2+ times • Depth of Repeat - The number of times consumers repeat purchase (1,2,3,4) • Adoption Rates - The rate of product adoption by consumers • Attractiveness - The attractiveness of the new product at the time of intro • Purchase Cycle - The time between 2 category purchases • Purchase Freq. - The number of purchases in a time period • Transaction size - The amount of product purchased for a purchase transaction (volume per purchase occasion) Concepts 3
- 4. 4 Accuracy and level of diagnostics should drive the choice of techniques. Never under-estimate “repeat” problems. Model Key Inputs Key Outputs Strength Weakness Parfitt and Collins Trial (T) Volume Easy Doesn’t capture Repeat (R ) What-if price/spending Buying Rate (B) Diagnostic Fourt-Woodlock CumeTrial (P) Volume by periods Easy Doesn’t capture Repeat Build (r ) Repeat/Trial sales What-if price/spending Depth of repeat Diagnostic Market-size units Assessor Awareness Volume Adoption/ Cost/Timing Trial by user group life-cycle of Category Buying HHs Innovations Structure/Hendry Market-size/structure Share, volume, profits Optimizing Judgment Share, price, spending Yr 1 and Going Yr launch on structure New Product marketing plan Cannibalization mgmt.. entry point Bases Purchase Interest Volume by year Validations Cost/Timing Transaction Size Best on near in Need for Norm Purchase Frequency line extensions database New Product Models - Summary 4
- 5. The source of inputs is predominantly via surveys (except Hendry) and should be behaviorally validated as much as possible. Model Key Inputs Sources Parfitt and Collins Trial (T) Past new product histories Repeat (R ) + Surveys on purchase intent, etc. Buying Rate (B) + Judgment Fourt-Woodlock CumeTrial (P) Past new product histories Repeat Build (r ) + Surveys on purchase intent, etc. Depth of repeat + Judgment Market-size units Assessor Awareness Category norms Trial by user group + Surveys on purchase intent, etc. Category Buying HHs + Internal estimates on market size Structure/Hendry Market-size/structure Consumer switching analysis Share, price, spending + Company P&L’s, Volume/pricing/distr from IRI/Nielsen New Product marketing plan + Judgment on effectiveness Bases Purchase Interest Category norm data base from prior surveys Transaction Size + Surveys on purchase intent, repeat, etc. Purchase Frequency + Judgment on Market definition/size New Product Models - Summary 5
- 6. Most forecasting models are based on the key underlying consumer dynamics driving sales Brand Volume = # of Brand buyers X Volume per buyer Current buyers # Purchases per buyer + X New Buyers Volume per purchase Concepts These behavioral measures are readily available from syndicated panel data 6
- 7. Parfitt and Collins conceptualized a simple model that has a great deal of intuitive appeal and has greatly influenced the structure and development of other new product models • It predicts ultimate market share for new repeat-purchase consumer products using input data from consumer panels. • Although the model requires actual market data (which is expensive since it presumes that the new product is at least in a test market), its ability to predict national share prior to national distribution can help management avoid future losses • Cumulative penetration (the total number trying the brand, over time) and repeat purchasing rates over time from the time each buyer first bought the product (along with a buying-rate index) form the basis for predictions of future share Parfitt & Collins 7
- 8. Parfitt and Collins predicts the brand share from three purchase measures: Trial, Repeat, Buying Rate Share = % Trial x % Repeat x Buying rate where – T = Projected percentage of triers of the new brand, – R = Projected percentage of those who tried and will repeat purchase the brand, and – B = Buying-level index of repeat purchase of the new brand, compared with an index of 1.0 for the product class average. Parfitt & Collins Example Share = % Trial x % Repeat x Buying Rate 8.5 = (34) x (0.25) x (1.0) That is, if 34 percent of the potential market tries this new product and 25 percent of the triers repeat purchase it, and they buy neither more nor less than other brands in the product class, the share for the new product will settle at 8.5 percent. 8
- 9. • A diffusion model which uses trial and depth of repeat • Estimates brand volume as the simple product of – Cumulative trial/penetration – Repeat purchase percentage rates for 1,2,3,4,5 repeats – Buying Rate • Only two values are necessary for the equation: – P = New Product Cumulative Trial at the end of the period – r = Rate of penetration for buyers who have not yet tried • Provides sales estimates by period (months or weeks) Fourt-Woodlock Fourt-Woodlock predicts the brand share from three purchase measures: Trial, Repeat rates and Depth if Repeat, Buying Rate 9
- 10. Study Design • One-cell Assessor® Positioning Evaluation • Mall-Intercept Screening – Males and females 18+ years of age – Meet past participation and security requirements • Central Location Interview – Purchase patterns for general product category – Preference and purchase behavior among consumers’ considered set of products – Exposure to and evaluation of new product proposition – Simulated purchase opportunity – Product/user perceptions Assessor 10
- 11. Sample Composition Three segments of respondents 1. Newbie 2. 2nd Adopter 3. Non-Intender • Six geographically dispersed markets Los Angeles Denver Atlanta Boston Chicago Dallas Assessor 11
- 12. The New Product Forecast Model • A share model that integrates the key principles and specific levels of - 1. Share/Loyalty of the category (e.g., 70 share brand has little room for growth, unless advertising grows the category). Also a new product potential is impacted by the current level of brand preferences within the category 2. Switchers available in the category (fewer the switchers, lower new product share potential and ability to grow more switchers, higher new product share potential) 3. Relative gaps on pricing and spending versus competition (e.g., greater the price differential to category average, higher the price sensitivity) 4. Position in the market structure hierarchy (lower in the structure, smaller the draw from the broader market) 5. Product effectiveness • Validated and extensively used by CPG companies (Kraft, Coca Cola, P&G, Unilever, etc.) and other consumer hard goods category’s (TV’s, Auto’s, Home Appliance’s, Electronics, etc.) Market Structure/Hendry 12
- 13. The simulation system can be used to determine how vulnerable a market is to new products and line-extensions • New Product Forecasting using Par Share Analysis – Determines the share of market and profit potentials of new brands, types, forms, flavors, package sizes, etc. under conditions of parity product quality. – The market vulnerability is estimated • Competitive Simulations and Gaming – Uses Advertising, Pricing, Par Share, Product Sampling, and Distribution Analyses to determine market vulnerabilities and opportunities for all competitors • Requirements – Competitive set, product effectiveness, year 1 marketing plan Market Structure/Hendry 13
- 14. Trial and repeat forecasts can be developed which are used to track actual in market performance Trial and Repeat Curves 0% 20% 40% 60% 80% 100% 0 1 2 3 4 5 # of Purchase Cycles %ofBuyers Repeat Trial Market Structure/Hendry 14
- 15. Key Assumptions - Cruncher Marketing Assumptions 2000 2001 2002 Snickers-Cruncher Total Working Media $ $0.0 $20.7 $10.2 Price per Pound 5.13 Price per 1.6 oz bar 0.51 Dist – ACV 90 Snickers-Original Total Working Media $ $19.7 $6.6 $16.9 Price per Pound 3.64 3.63 Price per 1.6 oz bar 0.36 0.36 Dist – ACV 99 99 Total Crispy Crunchy Segment (average brand) Price per Pound 3.89 Price per 1.6 oz bar 0.39 Dist – ACV 95 Market Structure/Hendry 15
- 16. Cruncher Simulation Results – back casting validation The simulation results were virtually identical to the 2002 actual. Other entry points (Line extension of Snickers, New brand of all Versatile) led to forecasts that were ~ 50 % lower than actual. Forecast Actual % Difference Consumption 3.88 3.87 - 0.3 % Share 0.813 0.810 - 0.3 % Market Structure/Hendry 16
- 17. The sales drop off in year 2 is consistent with new products trial build in year 1 Cruncher All Outlet Consumption 2001 2002 % Change Actual 4.52 3.88 (14) % Forecast 4.55 3.87 (15) % Market Structure/Hendry 17
- 18. Cruncher Source of Business Cruncher sourced disproportionably from the Crispy Crunchy Segment for the simulations Within Total Every Day Only (excludes Seasonal) Shipments Shipments Interaction Volume % source Index Cruncher 6.6 AO Crispy Crunchy-x Cruncher -1.7 -25% 156 Mainstream -3.3 -50% 99 Specialty 0.0 0% 0 Ver.Bites -1.7 -25% 88 Total 0.0 Within Mainstream - Snickers Original -0.6 -18% 110 - All Other Mainstream -2.7 -32% 100 Market Structure/Hendry 18
- 19. Share sourcing index back-up Final Share Sourcing Pre-Share Post-Share Share Chg% Shr Source Index Crispy Crunchy 16.0 16.6 0.6 Cruncher 0.80 0.8 AO Crispy Crunchy-x Cruncher 16.0 15.8 -0.2 25.0 156 Mainstream 50.7 50.4 -0.4 50.0 99 Specialty 4.9 4.9 0.0 0.0 0 Ver.Bites 28.4 28.2 -0.2 25.0 88 Total 116.0 -0.8 100.0 Market Structure/Hendry 19
- 20. The Ad response to new products is greater than established products New Product Ad Spending (mm) Share Ad Spending Impact on Long Term Share 6.8 5.8 Established Product Market Structure/Hendry 20
- 21. The higher spend in year 1 has a permanent long term impact on the new products ongoing share New Product Ongoing share with Higher Spending = 6.8 share Ad Spending (mm) Share Ongoing share with Parity Spending = 5.8 share Low $ High $ Ad Spending Impact on Long Term Share 6.8 5.8 Par @ $23 Intro @ $44 Market Structure/Hendry 21
- 22. There are 3 general Bases models for pre-market forecasting New Products/New Line Extensions Pre-BASES For screening concepts/positioning alternatives BASES I For concept evaluation BASES II For concept/product evaluation Bases 22
- 23. Bases also uses key assumptions about consumer purchase behavior to develop a volume forecast An advantage of the BASES concept screening system is that the survey integrates all of the key volumetric consumer measures, including measures for purchase interest, value perception, transaction size, and purchase frequency. The volume model integrates these measures into a single measure that can then be compared across concepts to assess their relative volume potential. - Purchase Interest - Transaction Size - Purchase Frequency Bases 23
- 24. Bases provides key measures for forecasting and planning • The objectives of a Bases study is to: – Understand the appeal of the concept/product proposition • Provide a comparison of the concepts to the BASES norm base of previously tested product concepts on the key measures of purchase intent, liking, uniqueness and value for money – Estimate unit sales and source of volume for the new product • Note that Bases results are based on forced exposure, which makes consumer awareness 100 % as well as availability (or distribution) 100 %, which most be adjusted for when forecasting Bases 24
- 25. Concept Key Meaures Quattro BASES' Database Comparison1 (N) Respondents (334) Tertile Purchase Intent (%) Top-Two Box 34 Bottom Definitely Would Buy 9 Bottom Probably Would Buy 25 Might or Might Not Buy 37 Probably Would Not Buy 17 Definitely Would Not Buy 12 Liking Rating (6 pt. Scale) 3.9 Bottom Price/Value (5 pt. Scale) 2.9 Bottom New and Different (5 pt. Scale) 3.2 Middle 1Compared to BASES' database of similar category concepts Database comparisons reported in thirds (bottom, middle, top) Bases provides key measures for forecasting and planning Bases 25
- 26. Higher Spend Lower Spend Total Year 1 Spending Households (M) 106,500 Year 1 Trial Rates 50% Repeats Per Repeater Average Repeat Units Marginal Product Performance Higher Spend Lower Spend Strong Product Performance $95 $75$95$75 Advertising $60 $40$60$40 106,500 106,500 106,500 Trial 40% 60% 48% 2.0 2.0 2.2 2.2 1.1 1.1 1.1 1.1 Packages (MM) 15.9 Volume (MM)* Year 1 Volume 12.7 18.0 14.5 39.6 31.7 44.9 36.0 Retail Dollars (MM) $149.1 $119.6 $169.1 $135.7 * volume does not include pipeline inventory. Bases can provide marketing spending plan simulations Bases 26
- 27. How appealing to the new product? Total Users Trademark Users (143) (57) Top-Two Box 53% 61% Extremely Appealing 29 33 Very Appealing 24 28 Somewhat Appealing 23 19 Slightly Appealing 10 9 Not At All Appealing 14 10 Bases can provide product and concept appeal scores Bases 27
- 28. Higher Spend Lower Spend Total Year I Spending (MM) $95.0 $71.0 Advertising (MM) $60.0 $40.0 Consumer Promotions (MM) $15.0 $11.0 Trade Promotions (MM) $20.0 $20.0 Maximum Distribution Food ($2MM+) 82% 82% Drug 88% 88% Mass Merchandise 100% 100% Other (Club) 40% 40% Trade Promotion Level Above Average Above Average Consumer Promotion Level Above Average Above Average Marketing Plan Scenario’s Bases can provide detail plan simulations Bases 28