How New Versions of Products Spread Differently Than Entirely New Products

Predictions about how products spread are usually based on the assumption that a few early adopters encourage an increasing number of people to start using the product. This assumption has largely been true of new technologies. But our research on phones, cars, and apps shows that when a product is not new but instead updates or replaces an existing product (which we call replacement innovations), the growth curve is very different, and managers making estimates around exponential growth are setting unrealistic expectations.

A better formula for estimating the early growth of replacement innovations follows a power law, with rapid adoption in the beginning followed by much more gradual takeoff as users make individual adoption decisions. This model generates more accurate predictions about adoption and, in turn, more realistic expectations and better understanding of resource needs.

New innovative products or technologies rarely emerge from a vacuum; in many cases they replace old technologies or earlier offerings that served a similar purpose. Consider 4K TV. The recently released technology delivers better image resolution for an enhanced TV-viewing experience. While that’s desirable, most people already have a TV — or three. So in order for them to buy a 4K TV, their calculus isn’t about just the new TV in and of itself; it’s also about whether to discard the old to bring in the new, at a somewhat hefty price tag.

To study the adoption of replacement innovations, we examined four areas: mobile phones (patterns for use of 885 handsets among 3.6 million Northern European customers between 2006 and 2014), cars (sales patterns for 126 automobiles sold in North America between 2010 and 2016), apps (daily downloads for the 2,672 most popular iOS apps from November 2016 to December 2016), and scientists’ research focus (246,630 scientists who published in 6,399 research fields). In each of these areas we documented the early growth of replacement innovations following a power law with non-integer exponents. This means that when the product was introduced, it had a singular growth momentum that was fundamentally different from its growth in the rest of sales periods.

To understand this, consider what happens when Apple releases a new iPhone. There may well be lines of early adopters outside the store for the device’s release, but subsequently, even if positive word-of-mouth spreads, the next waves of adopters will have to decide that they’re ready to replace their current phones with the new one. The slower, more deliberate decisions they make help to explain why replacement innovations grow more slowly.

We identified three mechanisms that are primarily responsible for the observed replacement dynamics: (1) recency, or how recently the new innovation has been introduced; (2) replacement propensity, as some products are more “fit” to replace original versions than others are; and (3) popularity, since more successful, or popular, products, are more likely to attract more new users — success begets success. A model that combines these three mechanisms enables us to explain the growth patterns of replacement products and to identify three parameters associated with this growth. These parameters are: fitness (how fit your product is to replace others), anticipation (initial excitement among potential users), and longevity (how long before the product may become obsolete). To understand intuitively how sales will go, ask yourself about each of these variables. The more recent or popular the innovation, or the better the product-market fit, the higher the sales.

Business leaders are often faced with the challenge of understanding which factors best determine whether and when a new product will succeed in unseating incumbent innovations. Our model offers a new way to start estimating the three adoption parameters once initial sales data becomes available. Decision makers can use these early figures and other related signals to assess whether the product’s fit, initial excitement, and shelf life are meeting expectations, and adjust tactics accordingly. For example, it may be important to focus on fit and improving product shelf life (longevity) during the design phase, with more attention to generating excitement (anticipation) among target users as the product launch nears.

For managers making decisions about when and how to release replacement products and innovations, this finding means that if you’ve been applying the traditional adoption model to what could be considered replacement products, you’re likely underestimating the initial excitement about your new product while overestimating the overall speed and size of adoption. This could mean wasted opportunities at the outset, followed by unrealistic expectations and, potentially, misallocated resources for the expected growth phase. On the other hand, our finding about the adoption of replacement innovations can help to create better prediction models, which may lead to better long-term organizational performance and create a significant advantage that competitors using other models can’t match.