Concept · Growth

Diffusion of innovation

Model how adoption spreads through a market, and separate adoption timing from product value

Adoption unfolds across people and time.

Diffusion describes how adoption of an innovation spreads through a population over time. The population and the adoption event must be defined: people who have ever bought the product, organizations that have implemented a practice, or households with active subscriptions are different measures. Adoption depends on awareness, access, affordability, compatibility with existing behavior, and perceived usefulness, as well as influence from earlier adopters.

In one sentence

Diffusion of innovation describes the process and timing by which a new idea or product moves through a defined population of potential adopters.

The Bass model concerns the timing of initial purchases. In a continuous-time formulation, dN/dt = [p + q(N/m)](m − N), where N is cumulative adoption, m is the defined market potential, and p and q are coefficients with inverse-time units. The right-hand side is an instantaneous rate, not automatically the adoption accumulated over a whole reporting period. A period total requires integrating the rate over that interval or using an explicitly specified discrete model. No numerical parameters or fictional adoption observations are assumed.

The original Bass paper tested a model against eleven consumer durable categories. Its structure assumes a defined market potential and initial purchase process; actual markets can change, repeat purchases can matter, and channels or product versions can reshape access. Diffusion is distinct from virality: viral growth concerns acquisition generated by existing users, while diffusion is the broader adoption path, which can be driven by advertising, distribution, regulation, or imitation.

Forces that shape the curve

Find the friction and influence affecting the next potential adopter.

External reach

Awareness, sales, distribution, or policy introduces the product to potential adopters.

Peer influence

Existing adopters share information, demonstrate use, or change perceived value.

Adoption friction

Price, access, compatibility, setup, or risk limits conversion and continued use.

A continuum, not a switch

Diffusion measures how adoption progresses over time within a defined population. Growth shape alone does not identify cause, retention, or profitability.

LowLimited adoptionHighBroad population adoption

“A rising adoption curve is a pattern to explain, not proof that the product improves with every user.”

Why it matters

A diffusion model can organize a capacity, coverage or communication decision around a defined adoption event. Fit it to consistently defined observations, hold back evidence for checking forecasts, and compare plausible explanations. A fitted curve can describe a trajectory without identifying whether awareness, peer influence, price, supply or distribution caused it. The mathematical relationship does not fill a missing market-potential estimate.

Apple announced that it sold its one millionth iPhone 74 days after the 2007 launch. This is a dated cumulative sales milestone and an observation of early purchase volume. One milestone cannot establish market potential, the adoption mechanism, the shape of diffusion, or the product's eventual economics.

Real-world examples

The same concept shows up in different ways across industries.

AppleAn early cumulative-sales milestone

→

Apple’s historical announcement reports the millionth iPhone sale 74 days after introduction. This is a dated cumulative-sales milestone, not an observed curve of unique people adopting through peer influence. The unit is the product sale under the company’s announcement; the total does not establish repeat purchases, the eventual market potential or the reason a buyer adopted. The analytical decision is what a diffusion model is meant to explain or forecast. Awareness, access, affordability and an appealing offer can change sales alongside imitation. A launch milestone cannot distinguish those explanations or identify the Bass model’s parameters. A consistently defined series and a justified eligible population would be needed for that calculation; no fitted curve or fabricated coefficients are supplied here. The company’s actual launch and observed milestone ground the timing question, while the missing adoption mechanism remains explicit. Use the model as a specified hypothesis with a defined event and population, rather than draw an S-shaped line from a successful product’s name.

When it breaks

The market-potential estimate may be wrong or change after launch, and the model can confuse the effects of advertising, price cuts, supply constraints, network value, regulation, and word of mouth. Sales are not always unique adopters, particularly when replacements, multiple products, or household accounts are involved. A sales milestone can also reflect channel inventory or multiple units bought by one buyer. Before using sales as adoption, specify what is being counted and what additional evidence connects the two. The Apple announcement is retained as a bounded sales observation; no diffusion coefficients are estimated from it.

Do not interpret a diffusion curve as proof of product-market fit or durable retention. A product can accumulate trial users while usage fades. Define adoption and active use separately, segment by channel and customer type, and test repeat use and contribution. When the market has multiple countries, network structures, or product versions, a single aggregate curve may hide materially different processes.

Key takeaways

01
What population and event count as adoption, and what time unit will be used consistently?
02
Which factors could explain the observed curve besides peer influence, and what data distinguishes them?
03
Are adopters retained and economically valuable, or does the curve measure only initial trial or shipment?

Sources

A New Product Growth Model for Consumer Durables · Management Science / INFORMS, 1969-01-01. Original printed pp. 216–217 (PDF pages 3–4): conditional purchase probability, cumulative initial purchases and instantaneous sales equation; body checked using in-memory PDF rasterization and OCR
Apple Sells One Millionth iPhone · Apple Newsroom, 2007-09-10. September 10, 2007, opening paragraph