Diffusion model

Frank Bass introduced the diffusion model in 1969 to forecast how a market adopts a new durable product. The model separates adoption driven independently by external influence from adoption encouraged by prior users, producing the familiar rise, peak and decline of new adopters over time.

When to use it

Use the Bass model when forecasting adoption, sales or installed-base growth for a genuinely new product, technology or service with limited historical demand data. It can support production, capacity and launch planning.

Price, advertising, geographic sequencing and launch phasing can shift the timing and parameters of adoption. The model’s characteristic curve is robust, but it should not be treated as immune to management action or market disruption.

The distinction between innovators and imitators explains why early external influence can start adoption and social or market interaction can accelerate it. This is compatible with Everett Rogers’s adopter categories: innovators, early adopters, early majority, late majority and laggards.

Origins

Frank M. Bass published the model in 1969 in his Management Science article “A New Product Growth Model for Consumer Durables.” He combined independent adoption and imitation in one parsimonious mathematical structure. Rogers’s earlier diffusion research provided important behavioural context, but the Bass model is a distinct forecasting equation.

What it is

Assume a fixed pool of potential adopters. Some adopt because of external influence such as advertising, distribution or independent interest; the model represents this with the innovation coefficient p. Others become more likely to adopt as the installed base grows; the imitation coefficient q represents that internal influence. The interaction produces an S-shaped cumulative-adoption curve and a single-peaked flow of new adopters.

The model does not directly classify people as permanently innovative or imitative. Its coefficients describe aggregate adoption dynamics for a particular innovation in a particular market.

How to use it

Let X be the market potential and Y the cumulative number of adopters. The remaining market is X − Y. The probability of adoption combines independent influence, p, with imitation proportional to prior adoption, qY. In the simplified form used here, new adopters N are:

N = (p + qY) (X – Y)

where

Estimate market potential and coefficients from analogous products, early sales or nonlinear fitting. Compare alternative assumptions, update parameters as observations arrive and evaluate error on later periods. Be explicit about units and time intervals.

Final analysis

The model has been highly influential in marketing and management science because it converts a plausible social process into a tractable sales forecast. Its usefulness does not mean every launch follows an identical curve; replacements, repeat purchases, supply constraints, competing standards and changing market potential may require extensions.

Online communities and networked products have renewed interest in diffusion forecasting, but direct network effects do not automatically make the basic aggregate model sufficient. Use it as a disciplined baseline and test richer models when the mechanism demands them.

Top practical tip

Treat market potential as an assumption to test, not a fact. Re-estimate p, q and the potential market as real adoption data arrive, and show decision makers how sensitive capacity or revenue plans are to each input.

Top pitfall

Do not confuse the aggregate innovation and imitation effects with fixed personality types. Adoption depends on the offer, price, infrastructure, communication network and social context, and the market potential may change over time.

Further reading

Bass, F.M. (1969) ‘A new product growth model for consumer durables‘. Management Science 15(5), 215–227.

Rogers, E.M. (2003) Diffusion of Innovations, 5th edn. New York: The Free Press