Marketing Plan Writing

Two market forecast models and the product life cycle

written by Tim Berry of Palo Alto Software


The Idea Adoption Model

The use of a model may be a good resource for creating a market forecast. Social scientists have studied adoption of new ideas by many different groups. They have studied how a group of doctors adopt a new treatment technique, how farmers accept a new farming idea, and many other examples of the process of groups of people absorbing innovation.

Most of these studies show a common pattern in adoption of new ideas. They are accepted first by the innovators, who seldom make up more than 2-3 percent of the larger group. The innovators prove the idea works, but they have relatively little influence on the group as a whole. The idea takes off when it is taken up by the early adopters (who tend to be the opinion leaders) who are a second group of roughly 13 percent of the larger group. These are people smart enough to see what those innovators are doing, and influential enough to spread that idea into the next group, the early majority, which comprises approximately 24 percent of the larger group. This pattern continues through the remaining groups, as the bell chart shows below, with the percentages in reverse as the number of people reached increases.

Idea Adoption

When you’re forecasting your market, particularly if you are looking at a new product and a new idea, the idea adoption model can help you improve your educated guessing. It’s not magic, not even strictly mathematical, but it can help.

Adoption Model-Examples

Calculate what percent penetration your market has at present, how long the market has taken to get there, and extrapolate based on the adoption model to calculate the rest of the curve.

Example: If the Internet took six months to spread to three percent of the market (Innovators), you might expect it to reach 16 percent of the market in another six months (Innovators and Early Adopters), and 50 percent of the market in another six months (Innovators, Early Adopters, and Early Majority). That would be an extremely fast ramp-up, but it still might be valuable to help you estimate.

Another example: Look at another technology that took a long time to spread. If only three percent of the potential market is using it after 10 years, then it might take another 10 years to reach 16 percent of the market.

You have to be careful how you apply the basic idea to your market. If you define your potential market as 12 million people and after five years only 200,000 own this new type of product, then the product doesn’t appeal to nearly as many people as you imagined, and you may have overestimated the potential market.

The Diffusion Model

You can model the spread of a product or idea on the spread of a disease, passed from one person to another. This is called diffusion. It can help you forecast a market.

Diffusion models were first used by health organizations to understand and predict the spread of contagious diseases. Market forecasters use them to simulate the spread of ideas, products, and techniques through groups of people.

The formula that follows is taken from James G. March’s published work, An Introduction to Models in the Social Sciences, which includes research into the use of diffusion models to market questions. It assumes that for a given population of size N, there is a diffusion factor a, which, when n is the number of people who already have the disease, the change in n during a time period is equal to the following formula:


Example: If the total population N is 50,000, and there are 5,000 people n who already have the disease, the formula would look like this:


Diffusion Model-Examples

In the early 1980s I headed a consulting group asked by Apple Computer to analyze the spread of personal computers into the Latin American market. We used a diffusion model to do it. This is a good example of practical use of diffusion calculations.

Our analysis started with estimates of the U.S. population of knowledge workers. Apple had contracted an earlier study on this and had given us the estimate of 50 million. Using standard idea adoption research, as outlined earlier, we redefined our total population into the classifications of 1) Innovators, 2) Early Adopters, 3) Early Majority, and 4) Late Majority. The table below  shows the breakdown for these diffusion group categories. Our forecast ignored the eight million so-called laggards and late adopters, which gave us a target market population of 42 million. In this case the diffusion factors were estimates.

  A B C D
1 Population Numbers Assumed Diffusion Factors
2     Factor 1 Factor 2
3 Innovators 1.5 Mil 1.50 n.a.
4 Early Adopters 6.5 Mil 1.00 0.50
5 Early Majority 17 Mil 0.60 0.30
6 Late Majority 17 Mil 0.20 0.10
7 Total Pop. 42.0 Mil    

The standard idea adoption research gave us a breakdown of the population into various adoption groups, and from there we developed diffusion factors for each group. We didn’t have primary research on diffusion, but we did have data on the early spread, and we manipulated the assumed factors to match the first six years’ market data we already had. The next table shows the early years of the forecast:

  A B C D E F G H
9   ’76 ’77 ’78 ’79 ’80 ’81 ’82
10 Users (thousands)              
11 Innovators 5 12 31 77 186 430 890
12 Early Adopters 3 11 38 114 317 822
13 Early Majority 1 5 19 64 197
14 Late Majority 1 3 9
15 Total 5 15 43 119 319 813 1,918
10 New Users              
11 Innovators 5 7 19 46 109 244 460
12 Early Adopters 3 8 27 76 203 505
13 Early Majority 1 4 14 45 133
14 Late Majority 1 2 6
15 Total 5 10 28 77 200 494 1,104

For spreadsheet experts only–you don’t have to follow this–the selected cell on the spreadsheet, E12, has the formula:


The related ranges are shown in Figure 23-7 for the cell references, and Early_Adopters is a spreadsheet range name for the 6.5 million early adopters included in the model.

The next table shows the breakdown for these diffusion group categories.

The chart below  follows this model up through 1999. We see an interesting picture of the spread of personal computers through the population of U.S. knowledge workers, as it might have been. This isn’t history, it is a market estimate using the diffusion model.

Diffusion Model One

What’s interesting, however, is how closely this matches the real market data as we know it. The curve shows an increase in computer usage among knowledge workers, reaching 40 million in the middle 1990s. It also shows that the knowledge workers who were new to computing in any given year reached its maximum in about 1988. By that time most of the market penetration was over.

This isn’t a complete analysis of market units because it doesn’t include replacement units and the second and third computers of the knowledge workers. Some of them have desktop computers at their offices and in their homes, and some have laptops. The replacement and add-on unit market drives the forecast after the middle 1980s.

As a second example, consider the spread of Internet use during the 1990s, beginning in 1994. In this case, the spread was much quicker and the population was greater, so we revised our assumptions as shown in the table below. We still ignored the laggards, but we made the total population 60 million. The diffusion factors are greater because the spread was faster.

  A B C D
1 Population Numbers Assumed Diffusion Factors
2     Factor 1 Factor 2
3 Innovators 1.8 Mil 8.00 n.a.
4 Early Adopters 7.8  Mil 4.00 8.00
5 Early Majority 20.4 Mil 2.50 4.00
6 Late Majority 20.4  Mil 2.00 2.50
7 Total Pop. 60.0 Mil    

The standard idea adoption research gave us a breakdown of the population into various adoption groups, and from there we developed diffusion factors for each group. We didn’t have primary research on diffusion, but we did have data on the early spread, and we manipulated the assumed factors to match the first six years’ market data we already had. The next table shows the early years of the forecast:

  A B C D E F G
9   ’94 ’95 ’96 ’97 ’98 ’99
10 Users (thousands)            
11 Innovators 5 45 395 1,800 1,800 1,800
12 Early Adopters 40 556 5,558 7,800 7,800
13 Early Majority 160 2,765 20,400 20,400
14 Late Majority 400 7,962 20,400
15 Total 5 85 1,111 10,524 37,962 50,400
10 New Users            
11 Innovators 5 40 350 1,405
12 Early Adopters 40 516 5,002 2,242
13 Early Majority 160 2,605 17,635
14 Late Majority 400 7,562 12,438
15 Total 5 80 1,027 9,412 27,438 12,438

This table above  shows the resulting data as if it were a forecast.  The chart below  follows this model through the periods 1994 through 1999. We can see how the idea spread more quickly through the population.

Diffusion Model Two

In this case we have over 40 million of the total 60 million knowledge workers already operating on the Internet by 1999. The same diffusion that took more than 20 years for personal computing took only five years for the Internet. The mathematics are the same, but the assumptions changed.

You will probably also notice in the chart that the basic phenomenon is the same. The overall market of users shoots up in an S-curve, but the new users reaches a peak and then trails off as the idea spreads through the population.

Product Life Cycle

Most products go through different stages of development, and those stages are reflected in the different growth rates expected at different stages. The important points are the takeoff point and the market saturation point.

The normal model of the product life cycle looks a lot like the same curve we use for tracking idea adoption, or the S-curve produced by the diffusion model.

Product Life Cycle

In the early development stage, growth rates may be high, but very few units are involved. This is a new market just beginning to reach its potential consumers’ minds and is even further from their pocketbooks. At this point, a few magazine articles begin to address this new technology. As Innovators begin to look for the product, the channels of distribution are set up.

Takeoff is when the market turns upward. Pioneer buyers have spread the word and the general public begins to buy. This is what happened when, for example, mall merchandisers began to sell a lot of home computers in 1982, or when color television sets took off in the 1960s. This is the stage all those people who invested in the early market are waiting for. Growth rates are still attractive, and volume has skyrocketed.

Growth rates decline when the market approaches saturation. At this point, most of the buyers who want the product have it. The market turns into a low-growth and replacement market, just as the markets for stoves and refrigerators did.

Using the Life Cycle in Forecasts

There are three keys to remember in using the product life cycle in your market forecasts:

  1. Takeoff resembles a snowball, as it rolls down the hill gaining momentum, and the situation changes enormously.
  2. Once markets get going, saturation will occur.
  3. Each product has a different life cycle, some fast, some slow, and some unlike the normal pattern.

Takeoff is the trickiest stage. There are markets that never take off at all. Some companies spend years waiting for the snowballing affect to get started, and it never does. There are also markets that take off at odd moments, at times you would not normally expect, or after long periods of apparent smoldering. For example, the home computer boom was predicted as early as 1978, but didn’t happen until 1982. Some companies went broke waiting for it. Others left this market and were not participating when the boom finally happened.

After takeoff, it’s too easy to forget that market growth rates will eventually approach saturation and decline. Some of the color television people projected increasing growth through the 1970s and were caught off guard when the market moved up to about 10-15 million units per year and then stopped. The video market drew a crowd of new entrants in 1982 and then virtually stopped growing.

The speed of a life cycle is also hard to predict. The video game industry boomed and faded like a typical fad. The home computer industry will probably last longer before saturating its market; then it will become a replacement market. Much depends on what the product does for its buyers and on who those buyers are. Sometimes the standard idea adaptation research will help.

Mature markets are by far the easiest to forecast. These have slow growth rates and little change from year to year. In these markets, the old-fashioned forecasting methods – such as taking the average growth rate of the last five years and projecting it into the next five years – work reasonably well.

Tim Berry

about the author

Tim Berry

Founder and President of Palo Alto Software and a renowned planning expert. He is listed in the index of "Fire in the Valley", by Swaine and Freiberger, the history of the personal computer industry. Tim contributes regularly to the bplans blog, the as well as his own blog, Planning, Startups, Stories. His full biography is available at Follow Tim onGoogle +


Getting ready to create
a marketing plan?

Get practical ideas and good models with dozens
of examples of successful marketing plans