📊 Innovation Diffusion — Bass Model

Bass diffusion model: dN/dt = (p + q·N/M)·(M−N). Innovators (p) and imitators (q) drive S-curve adoption. Fit historical data: TV, smartphones, EV. Find peak sales time t* = (ln q − ln p)/(p + q).

SocietyInteractive
Orange = new adopters (sales) · Green = cumulative adopters · Dashed = peak

How it Works

The Bass model (Frank Bass, 1969) is the canonical model of new product diffusion. It separates adopters into two types: innovators who adopt regardless of others (driven by advertising, curiosity) and imitators who adopt due to word-of-mouth from existing users.

The differential equation for new adopters per unit time is:

dN/dt = (p + q·N/M) · (M − N) N(t) = cumulative adopters at time t M = market potential (total eventual adopters) p = innovation coefficient (0.001–0.03 typical) q = imitation coefficient (0.3–0.5 typical) Peak sales time: t* = [ln(q) − ln(p)] / (p + q) Peak sales rate: dN/dt|_peak = M(p+q)²/(4q) Closed-form solution: N(t) = M · [1 − e^{−(p+q)t}] / [1 + (q/p)·e^{−(p+q)t}]

The ratio q/p governs the shape: large q/p produces a pronounced bell-shaped sales curve (strong social effect); small q/p gives monotonically declining sales (innovator-driven). The model is integrated numerically via Euler steps here.

Frequently Asked Questions

What is the Bass diffusion model?

The Bass model (1969) describes how a new product spreads through a population of potential adopters M. Adoption is driven by two forces: innovators who adopt independently (coefficient p) and imitators who adopt due to social influence from existing adopters (coefficient q).

What does the Bass model equation say?

dN/dt = (p + q·N/M)·(M−N), where N is cumulative adopters, M is market potential, p is the innovation coefficient, and q is the imitation coefficient. The first term represents innovators; the second represents imitators influenced by the installed base.

What is the peak sales time in the Bass model?

Peak sales occur at t* = (ln q − ln p) / (p + q). This formula shows that peak time decreases when q/p is larger (stronger word-of-mouth) or when p+q is larger (faster overall diffusion).

What is the S-curve in adoption?

The S-curve shows cumulative adopters starting slowly, then accelerating as imitators pile in, then decelerating as the market saturates. It appears across technology adoption: radio, TV, internet, smartphones.

How are Bass parameters estimated?

Parameters p, q, and M are typically estimated by fitting the model to historical sales data using ordinary least squares or nonlinear regression. Once estimated, the model forecasts future sales trajectories and helps identify peak time and total market size.

What real products follow Bass diffusion?

Frank Bass originally fit the model to color TV adoption. Subsequent studies validated it for refrigerators, air conditioners, computers, mobile phones, streaming services, and electric vehicles.

What is the difference between p and q parameters?

The innovation coefficient p represents adoption independent of social influence. The imitation coefficient q represents word-of-mouth. Typically q >> p; most products have p ≈ 0.01 and q ≈ 0.3–0.5.

What is market potential M?

Market potential M is the total number of eventual adopters. Estimating M is challenging as it depends on price trajectory, competing products, and demographic changes. M is often the most uncertain parameter.

Can the Bass model handle multiple products?

The Norton-Bass model handles successive technology generations (e.g., iPhone generations) and models competitive substitution. New generations cannibalize old ones while growing total market.

How does the Bass model relate to the SIR model?

The Bass model is mathematically related to the SIR epidemic model. Innovators are like external infection, imitators are like person-to-person transmission, and market saturation plays the role of population immunity. Both produce S-shaped curves.

About this simulation

This simulation numerically integrates Frank Bass's 1969 diffusion equation dN/dt = (p + q·N/M)·(M−N) with simple Euler steps, plotting both the green cumulative-adopters S-curve and the orange new-adopters-per-year bell curve. Switching to one of the real-world presets — color TV, PC, smartphone, or EV — loads historical p and q coefficients so you can compare how differently technologies actually diffused.

🔬 What it shows

A cumulative adoption curve N(t) rising from 0 to market potential M, an overlaid sales-rate curve dN/dt that rises then falls, and a dashed vertical marker at the computed peak-sales year t*.

🎮 How to use

Drag the innovation coefficient p, imitation coefficient q, market potential M, and time horizon sliders, or pick a preset product from the dropdown; switch the Display selector between cumulative-only, sales-only, or both curves at once.

💡 Did you know?

Frank Bass originally fit this exact equation to historical color television sales, and the same two-parameter model has since matched adoption curves for products as different as refrigerators, PCs, smartphones, and electric vehicles — all sharing the same underlying innovator/imitator mathematics as the SIR epidemic model.

Frequently asked questions

Why does the sales curve rise, peak, then fall while cumulative adoption keeps climbing?

New adopters per year (dN/dt) tracks the momentary rate of growth of the S-curve, so it necessarily peaks at the S-curve's steepest point and then declines toward zero as the market saturates near M — cumulative adopters N(t) simply keeps accumulating that shrinking flow.

Why do the presets have such different q/p ratios?

Products with strong word-of-mouth and network effects, like smartphones (q=0.60) or EVs (q=0.55), have much higher imitation coefficients than early color TV (q=0.25), reflecting how much more adoption today is driven by social influence rather than independent innovators.

What actually determines the peak-sales year shown in the stats panel?

It's computed directly from t* = (ln q − ln p)/(p + q); a larger p+q sum compresses the whole diffusion timeline, so products that spread faster overall — even with modest q/p ratios — hit their sales peak sooner, which the dashed marker tracks live as you move the sliders.

Why does raising market potential M not change the shape of the curves?

M scales the S-curve's ceiling and the sales curve's height but leaves the underlying dynamics (governed purely by p and q) unchanged — the simulation's y-axis labels rescale automatically, but the timing of the peak and the overall S-shape stay identical.

Is the Bass model just the logistic growth curve in disguise?

Not quite — pure logistic growth needs no innovators (p=0) and grows purely from existing adopters converting others; the Bass model's extra p term lets diffusion start even with zero initial adopters, which is exactly why it can model brand-new product launches where logistic growth alone would predict zero sales forever.