Background

New products can help improve penetration by making portfolios more relevant to more category buyers (see Ehrenberg-Bass Sponsor report #76), and new products can even help brands to charge higher average prices. But, of course, there is risk when launching a new product that it will fail to generate enough mental and physical availability, or even be rejected by consumers. This risk, coupled with the time and financial resources needed to develop and support new products at launch and beyond, makes new product development daunting.

There is a lingering perception, found in both academic and industry articles (e.g. Gourville, 2006; IRI, 2016), that the majority of new products, sometimes even nine in ten, will fail. These claims need empirical validation. It is important to have realistic new product failure rate expectations, because if managers think they are higher than they really are, they may dismiss their failures as being normal/understandable and not look hard enough to learn from their mistakes. We examine whether the belief that most new products fail is empirically grounded or an unfounded marketing myth. We also examine the category and brand conditions under which new product failure rates can vary.

Past research

First, we assembled prior evidence about new product fail rates. We identified several past reviews that conclude failure rates are lower than often believed (e.g. Castellion & Markham, 2013; Cierpicki et al., 2000; Crawford, 1979; Crawford, 1987). We also identified eight peer-reviewed studies over the past 26 years that report new product failure rates. Across these eight studies, between 29 to 46% of new products fail (Table 1). While the methods, samples, data, and measures vary, no empirical studies report a failure rate higher than 46%. Most studies use firm judged measures and report failure rates around 40%. One study used an objective measure (ceasing to sell after one year) and listed failure close to 30% (Wilbur & Farris, 2014). We use this study as a basis for our investigation to identify the conditions under which failure rates vary.

Table 1: New product failure rates (peer-reviewed studies published in the last 26 years)

Our investigation

We examined failure rates of over 30,000 new products (defined as new SKUs) launched in the US either in 2012 or 2013, across 25 packaged goods categories (from soft drinks to bath tissue). We investigated all new products launched by big and small, declining and growing brands, and from stable, declining and growing categories (Table 2). We identified failure as a new product ceasing to be bought (no sales) US wide, one and two years after launch.

Using this criteria, we found the average failure rate is 17% one year from launch. Over two years, it rises to 37% (Figure 1). Our failure rate figure at two years is quite similar to the results from past empirical research cited in Table 1.

Figure 1: Percentage of new products failing from launch across categories

Important conditions

Investigating failure across categories is an important first step, but it is also important to examine the category and brand conditions where we expect failure rates could vary: (1) Category size and dynamics; (2) Brand size and dynamics; and (3) New product types.

1) Category size and dynamics

Firstly, we found new product failure rates vary quite lot across the 25 categories. Failure can be as high as 57% (chocolate) and as low as 13% (ready to heat rice) two years from launch (Table 2). With the variation we see in Table 2, we examined whether failure rates vary predictably across categories with high to low yearly sales and across categories with different dynamics. We first compared new product failure rates from high to low dollar sales categories. This revealed that higher dollar sales categories have more new products and they also tend to have proportionally higher failure rates than in smaller categories (also see Table 3).

Table 2: Data profile and percentage of new products failing across categories

Next, we looked at growing and declining categories (i.e. categories that increase/decrease in dollar sales year-on-year). We use a similar approach to Sharp et al. (2017) (see Ehrenberg-Bass Sponsor report #74), where categories are classified as growing where annual dollar sales increases are greater than 5%. However, we classified declining categories as annual dollar sales decreases that are greater than 5%. All other categories were considered stable in our analysis.

Most categories we analysed were stable, and most launches were from stable categories (Table 3). Growing, stable and declining categories featured similar failure rates (all around 40%). We found that new products in declining categories have a harder battle to remain on shelf. However, this result is based on one category, so more robust evidence is needed to draw generalisable conclusions.

Table 3: Percentage of new products failing after two years from launch by category size/dynamics

2) Brand size and dynamics

Next, we examined new product failure rates for different market shares of brand and different growth trends. We used the same approach to classify brands that are growing, stable and declining in sales that we used for categories.

We discovered that larger brands launch more new products, a similar result to what we see for categories. Large and medium sized brands have proportionally lower failure rates (at 36 and 33%, respectively) compared to smaller brands (45%) (Table 4). This may be because larger brands likely have more resources to develop and support new launches. Being easy to find and buy on shelf is key, so an advantage larger brands have is the ability to build ‘presence’ in shopping environments (see Ehrenberg-Bass Sponsor report #76). However, the slightly higher success rate for larger brands may be because the new SKU gains sales, but, primarily, from other SKUs under the same brand. Previous Ehrenberg-Bass Institute research has shown that buyers of new product variants are more likely to be heavy buyers of the parent brand (Tanusondjaja et al., 2016). 

When comparing new product failure rates from growing, stable and declining brands, we found that growing brands collectively launch more new products (50% of new products were launched by growing brands). Unsurprisingly we also find growing brands have proportionally lower failure rates on average (34%), compared to stable or declining brands (41 and 42%, respectively) (Table 4). However, what is still to be determined is whether these brands grew because the launches were successes, or whether these new products are more likely to be successes because these brands were growing. 

Table 4: Percentage of new products failing after two years from launch by brand size/dynamics

3) New product types, e.g. new launches by new brands and existing brands

Finally, we wanted to look at new brands, compared to new products launched as part of an existing brand’s portfolio. New products under a new brand name are a lot less common compared to launches by existing brands, and they have a lower failure rate (Table 5). Approximately one in three products launched from a new brand are not sold two years later (30%). This is seven percentage points lower than the failure rate for launches from an existing brand (37%). A lower rate of success for brand extensions has been reported previously (Sharp 1991, 1993) with lower marketing support being blamed, probably because marketers expect/hope that their use of an existing brand name can give them a cheaper launch.

Table 5: Percentage of new products failing after two years from launch by new and existing brands

Summary

New products do fail but failure is not as high as is widely believed, or at the very least, not as fast as typically thought. Based on peer-reviewed empirical research from the past 26 years, and in our analysis of 30,000+ new products across 25 packaged goods categories in the US, we report that new product failure is 17% one year from launch and increases to 37% after two years. These average rates are much lower than the ‘eight in ten’ sort of figures that are often reported. 

Failure rates are higher in larger categories but lower for larger brands. They are also lowest in stable categories and for brands that are growing. Finally, we find failure rates for new products launched under a new brand name are lower than failure rates for new products launched under an existing brand name, in line with previous studies from decades ago.

Our findings help marketers understand what is ‘typical’ in new product failure rates. Given that failure rates are lower than expected, marketers should work harder to understand the reasons for their failures. Today, most launches are subject to a lengthy development process with considerable testing. The reasons for failure, like the reasons for extreme success, are therefore unlikely to be obvious (e.g. it tasted awful). We encourage marketers to dig deep, to investigate and learn from product failures.