New Product Sales Forecasting
Forecasting sales for a new product is fundamentally different from forecasting an existing one. There is little or no historical data, customer response is uncertain, and the business often needs a forecast that is not only accurate enough to guide decisions, but also meaningful enough to support launch planning, investment choices, and risk management. Academic reviews therefore treat new product forecasting as a multi-method discipline that spans judgment, market research, simulation, diffusion modelling, and newer data-driven approaches.
Our Methodology
Concept Testing, Purchase Intent, and Choice Modelling
Our methods uses direct customer input. This includes purchase-intent surveys, concept tests, and multiattribute methods such as conjoint analysis. The choice based-conjoint analysis approaches are especially valuable before launch because they allow businesses to estimate likely demand even when no sales history exists. They are most useful when the product can be described as a bundle of attributes, benefits, and price points, and when customer trade-offs are central to adoption. The main challenge is that stated intentions do not automatically translate into a full time-series forecast, so additional assumptions or modelling steps are often needed.
New Product Sales Forecasting for High-Tech Products
High-tech products require a more adaptive forecasting approach because customer preferences, product attributes, and competitive benchmarks can shift quickly. Earlier research found that successful high-tech industrial projects tended to rely more on internal qualitative forecasting techniques than low-tech projects, reflecting the higher uncertainty and faster change in these markets. At the same time, more recent work shows that formal quantitative models remain essential, particularly when they are refreshed with current data and combined with structured market research.
Choice-Based Conjoint Analysis for New Product Sales Forecasting
Choice-based conjoint analysis is especially valuable for high-tech products when adoption depends on trade-offs between features, performance, ecosystem, service, and price. In this method, customers are shown realistic product alternatives and asked which one they would choose. The model then estimates the utility attached to each feature and predicts the probability that a customer or segment would choose the new offer over competing options. Research shows that conjoint-based models can be used not only to estimate market share at a point in time, but also, when combined with assumptions about future prices and category diffusion, to generate time-series sales forecasts.
For high-tech categories, this method is powerful because it helps answer three critical launch questions:
-Which feature combination customers value most?
-How sensitive demand is to price and specification changes?
-How the new product is likely to perform against competing offers?
However, research shows an important caution: for high-tech, short-life-cycle products, the relative importance of attributes can change quickly over time. That means conjoint data should ideally be collected close to launch, or the model should be refreshed and updated as new behavioural data becomes available.
Implementatipon / Applications
The first approach is primary choice experimentation. Customers take part in a structured study and choose between different product configurations before launch. This is ideal for concept evaluation, pricing strategy, range architecture, and go/no-go decisions when the product is still being designed. The second approach is behavioural updating. Once the product is in market, real choices such as clicks, comparisons, basket additions, subscriptions, and purchases can be used as revealed-preference data to refine the original conjoint-based forecast. Research on Bayesian updating shows that combining stated-preference and revealed-preference data can improve confidence in forecasts for newly introduced technologies with limited sales history.
Further Reading
Our approach is grounded in established academic and industry research. To learn more about the methods and evidence behind this work, we recommend the following resources:
S Jahanbin, P Goodwin and S Meeran, "New Product Sales Forecasting in the Mobile Phone Industry: an evaluation of current methods", International Institute of Forecasters, 2013.
Here is the LINK
https://forecasters.org/wp-content/uploads/gravity_forms/7-2a51b93047891f1ec3608bdbd77ca58d/2013/07/Jahanbin_Semco_ISF-2013.pdf
S Meeran, S Jahanbin, P Goodwin and J Q F Neto, "When do changes in consumer preferences make forecasts from choice-based conjoint models unreliable?", European Journal of Operational Research, 2017.
Here is the LINK
https://purehost.bath.ac.uk/ws/portalfiles/portal/147016138/EJOR_paper_last_submission_which_was_accepted_19_08_16.pdf