1. Data-driven product forecasting
In this last video, we'll cover the Product aspect of marketing's 4Ps. You will use data to figure out how to navigate a new product launch.
2. Not always simple
You may think forecasting sales for a completely new product requires making up numbers. Year one we will sell 100 widgets and year 2 will have 20% growth, so 120 widgets and so on. However, there are data-driven models that can help you even when the product is completely new.
3. Products on products on products
I was lucky enough to work in operations at a large tech company that created new e-readers, a color tablet, an app store, and streaming videos all within three years. That's a lot of new products and services! Being in operations, my job was forecasting customer demand, ultimately leading to customer service phone calls.
4. Growth model example
There are multiple growth models for new product forecasting. Growth models forecast demand for new products, or in my case how many phone calls, emails and chats to expect.
One example of a growth model is the BASS model. Bass models require only a few data points and seek to mimic early and late consumer behaviors.
5. Innovation vs. imitation
The BASS model has three key inputs.
First, the total market or m parameter represents the total market capacity, exclusive of your company. This is a fixed assumption in your model for the total number of widgets, e-readers, phone calls, whatever you are forecasting in its entirety.
Next, the p and q parameters represent the rates of innovation and imitation respectively. The rate of innovation represents the people that buy the product early because of features and marketing. The rate of imitation is the rate of people buying based on word of mouth or reading reviews.
As the market unfolds of time, the total forecast is P+Q. However, the entire sum of P+Q over time cannot exceed the assumed total market.
6. P: Innovation behavior
Innovators decrease over time steeply. These people are willing to take chances early on in a product life cycle. Maybe they are enticed by a promotion or just excited for new technology or products overall. In total there are fewer innovators in a market compared to the late-adopters, or imitators.
7. Q: Imitation behavior
In contrast, imitators are late to the party. They represent a larger portion of the total market but are slower moving. Once imitation takes hold it accelerates quickly. Lots of people want to get the benefit of the product once they are comfortable with reviews or someone they trust tells them it's good. Keep in mind the total market is limited so a decline eventually occurs even among these imitators.
8. P & Q side by side
Often failed products still sell to early adopters enticed to take a risk but then there are not enough imitators to sustain a product. If the product gains traction, by word of mouth or online reviews, then imitators will start to make purchases. Often, companies add new features and update models to spur a new wave of innovator purchasing. Summing the P+Q for any specific point in time, like month 20, represents the market forecast for the time period.
9. Altogether now
This visual contains all BASS components in a more realistic visual. The blue line is the declining innovation rate, P. The green is the increasing imitation rate, Q. As the market starts, notice the blue line is above the green line until period 3. The sum of these two is the total market adoption for each period shown in red. Note how the red curve closely approximates the actual values represented as black dots. It's not perfect but with only three parameters this curve is better than a naive forecast.
10. Historical P & Q
Usually BASS modelers find similar historical products as a starting point. These are example of P and Q values where the outcome was known. Overall, average P and Q values are 0.03 and 0.38 respectively.
11. Let's practice!
Next, you'll use a BASS model to choose an analog product, then adjust it slightly to fit the limited sales data you already have. Let's get to it.