How to identify line extensions that survive

Many think innovation is critical for growth but few CEOs are satisfied with their organization’s innovation performance (McKinsey and Company, 2023). Although some consumer product launches generate millions in sales over their first year (e.g. Circana, 2024), this success is uncommon (e.g. Victory et al., 2024). This slim chance of success is concerning, given the high cost to develop and launch new products. This risk highlights the need for a greater understanding about how to improve the chances of achieving new product success. The context of this study is line extensions for consumer goods brands.

One of the strategies marketers use to encourage new product success is leveraging an established brand name and extending it to a new product. Extensions make up the vast majority of new launches (Kovalenko et al., 2022), perhaps because they are considered to be a “safer bet” than new brands. Extensions include an established brand introducing a new product in a category that the established brand has not previously operated in (brand extension) or launching a new product in a category that the established brand already operates in (line extension) (Aaker and Keller, 1990; Reddy et al., 1994; Tauber, 1981).

There are several reasons why leveraging an established brand name to introduce a new product is an attractive route for marketers. One key reason why extensions are more attractive is that extensions can transfer the parent brand’s associations in consumer memory to the new product (Aaker and Keller, 1990). Evidence suggests extensions can be beneficial as they can influence consumers’ attitudes and intentions (see, Peng et al., 2023) and encourage cross-purchasing across categories (e.g. Grasby et al., 2022). The latter is a positive outcome, particularly when extending brands across categories. Although excessive cross-purchasing within the same category (line extension) may not always result in poor results (e.g. cross-purchasing between low and high margin products), it might lead to a net negative outcome.

Despite the aim to introduce products that are incremental to the brand within the same category, cannibalization within a brand portfolio is the norm (Lomax and McWilliam, 2001). In addition to the concerns around cannibalization, supporting new line extensions (LEs) also has implications for resource allocation, often at the expense of other products within the brand's portfolio. As an example, retailers have finite space allocated to categories and choosing to allocate space to a LE will likely sacrifice existing facings of the brand’s core range. This can be damaging because a brand’s top selling product typically attracts around half of all brand buyers (Tanusondjaja et al., 2018).

Although a substantial sum will already have been invested pre-launch that cannot be recovered, given the potential cannibalization and the significant resources still required at launch for LEs, there is a need for marketers to better distinguish the introductions that have a greater chance at achieving success or identify likely failures early on. This would help marketers to either:

• ramp up the investment for a LE that is showing positive signs; or

• begin to curtail or adjust planned LE initiatives.

Monitoring success over a new product’s launch year is considered critical (Lamey et al., 2018; Sinapuelas et al., 2015; Victory et al., 2024) but there is little guidance on the early signs new launches exhibit when they are on a path to failure or success (e.g. Asplund and Sandin, 1999; Åstebro and Michela, 2005; Singh et al., 2012). This leads to the key question of this study: Can survivors be identified from their launch year performance?

To answer this question, an “empirics-first” (empirical-then-theoretical) approach is used (see, Ehrenberg, 1994; Golder et al., 2023; Sharp et al., 2017) to examine the long-term (three years) and first-year launch performance of thousands of LEs. This study uses a total of 12 years (2004–2016) of consumer panel data across five consumer goods categories from four United States regions to identify LEs, measure their survival over a three-year period and compare the quarterly performance of each LE during its first year after launch.

Overall, 7,195 survivor LEs (i.e. reported sales after three years) and 5,294 failed LEs (i.e. stopped reporting sales after the first year) were identified and their performance on two critical brand performance measures that constitute market share were compared. These two performance measures are:

1. repeat-buyer rate; and

2. penetration.

Adopting easily accessible performance measures in industry-relevant data is important because it can be difficult to get a read on a suite of metrics when products are new and not widely visible or known within the category, particularly metrics that are collected from surveying current or potential category buyers.

This research makes two main contributions to marketing practice. First, the findings highlight the need to focus on changes in penetration as a critical indicator for monitoring LE survival. This suggests that the launch strategy to improve LE survival should prioritize recruiting new buyers first and foremost, rather than focusing on activities to encourage loyalty among the triers. Second, the importance of penetration for long-term LE survival is evident as early as the first quarter after launch. Since early trial is a signal of potential survival, marketers should adjust marketing mix and investment levels accordingly to capitalize on the success, or remediate their actions if penetration levels are low.

All of the world’s most popular products were once new, but many new products do not survive. Around one in four new consumer products are no longer purchased one year after launch (e.g. Davies, 1994; Wilbur and Farris, 2014). A 25% chance of failure might not be particularly dire, but a product’s failure rate continues to rise the longer the time since launch (Anderson et al., 2015; Victory et al., 2021).

Extension strategies are often used to reduce the risk of failure. Past research shows brand extensions generally have greater market shares than new brands (Smith and Park, 1992). This is thought to occur because extensions “borrow” associations from their parent brands, including quality signals and brand familiarity (Pitta and Katsanis, 1995), while new brands effectively start at “zero.” The theory behind this mechanism is categorization theory, which posits that our memory categorizes information based on their similarity to facilitate retrieval (Mervis and Rosch, 1981). This benefit also extends to LEs, as LEs are more likely to be purchased by consumers with greater experience with the parent brand (e.g. Kim and Sullivan, 1998; Tanusondjaja et al., 2016). Although LEs might have an initial advantage in inducing trial, particularly from higher share brands (e.g. Sinapuelas et al., 2015), the long-term survival of LEs is never assured. For example, a recent study reports a LE failure rate (non-survival) nearing 80% three years after launch (Victory et al., 2024).

In addition to performance varying by product type, it also can vary depending on how success is measured. For example, practitioners estimate only around 40% of all new products launched over a five-year period will fail (e.g. Barczak et al., 2009; Knudsen et al., 2023; Markham and Lee, 2013). This is lower than the results from sales data which suggests practitioners may be overestimating new product success. Indeed, there is bias in the practitioner evaluation method because it relies on marketers to accurately recall launched products and judge their success. However, marketers are no better at judging the success of marketing interventions than a coin toss (e.g. Hartnett et al., 2016). Although that study was in an advertising context, it further highlights the need to use explicit measures (over practitioner evaluation) to pinpoint successes and prevent failures.

Although this evaluation method ensures that a range of success perspectives are captured, the diversity in the criteria used can lead to “measurement error and theoretical ambiguity” (Åstebro and Michela, 2005, p. 323). This ambiguity, along with past studies not being explicit about the types of new products examined, has created challenges in establishing generalized relationships about why some new products are far more successful than others. Concerns about relying on practitioner evaluation to measure new product success have also been highlighted in a recent study (see, Victory et al., 2024). In response, investigating early changes in practitioner-relevant, explicit success outcomes for LEs is the aim of this study.

Measuring and interpreting new product success can be extremely difficult. For example, the same new product can be considered “successful” on some success measures but not on others (e.g. Hultink et al., 1999). Furthermore, marketing practitioners might find it challenging to measure new product success because success may take longer to manifest in the data. Although there is currently no clear understanding about the metrics that might help predict new product success early on, there are already generalizable patterns in how two performance metrics vary, which explain why some brands sell more than others.

An abundance of research, spanning several decades, shows that a key difference between competing brands with higher and lower market share is in the number of customers they have (i.e. penetration), rather than how often their customers purchase them (i.e. loyalty) (Ehrenberg et al., 1990; Graham et al., 2017; Martin, 1973). This predictable pattern is called Double Jeopardy. Double Jeopardy is one of the most well-established empirical “laws” in marketing, as it has been shown for countless brands across a variety of conditions (Sharp et al., 2024). Double Jeopardy is a key building block of Dirichlet theory (Goodhardt et al., 1984). The lack of variability observed in behavioral loyalty (e.g. repeat rate, purchase frequency) is one of the assumptions of the NBD-Dirichlet model (Goodhardt et al., 1984).

The importance of building brand penetration has been recognized by prior empirical work investigating market share growth for established brands (e.g. Romaniuk et al., 2014, 2018). Collectively, many studies conclude that brand growth depends on prioritizing activities that are primarily aimed at recruiting new buyers to the brand, as continuing to attract non and light buyers is critical to success (Riebe et al., 2014; Trinh et al., 2024). Although the recommendation to recruit new buyers to grow established brands is well-known, its relevance to new product strategy (vs increasing loyalty) is not yet known.

There is considerable knowledge about the early performance patterns of new brands. New brands are shown to perform similarly to (comparably-sized) established brands early on, exhibiting the “as expected” patterns soon after launch (Ehrenberg and Goodhardt, 2000; Hoek et al., 2003). Another study shows new launches exhibit a similar pattern but they tend to have slightly lower loyalty than their (comparably-sized) established counterparts (Trinh et al., 2016). Together, this suggests that, at least for completely new brands, prioritizing activities to build penetration may be a more productive pathway to achieving new brand success.

Another study investigating the early performance of brand extensions found that the “average” (surviving) brand extension with a rising sales trajectory over a three-year period had higher penetration and loyalty in each quarter of the launch year compared to the “average” (surviving) brand extension with a declining sales trajectory (Singh et al., 2012). Along with the findings about completely new brands, these results suggest surviving new brands (including extensions) should prioritize building trial rather than loyalty during launch.

Although the above findings help marketers to identify (likely) new brand successes and failures early on, it remains unclear whether the findings will apply to other types of new products (which may vary in their success drivers). One study examined this question and found that longer surviving new beer products have higher market shares than new beer products that failed earlier (Asplund and Sandin, 1999) but the underlying mechanisms of market share (i.e. penetration and loyalty outcomes) were not investigated. Addressing this gap is the aim of the current study.

Recent research has begun to adopt more explicit metrics, such as survival time (e.g. Salnikova et al., 2019, 2020; Victory et al., 2021, 2024; Wilbur and Farris, 2014), to measure new product success. A recent study (see, Victory et al., 2024) benchmarking the adoption of LEs explored the likelihood of survival over the first few years but it did not investigate the specific performance indicators that could be used by marketers to predict long-term survival. Obtaining an early read on the likelihood of survival is important to help marketers decide whether to continue support.

There are multiple ways to measure new product success, including survival, failure and kill rates, profit, sales or relative contribution, or a qualitative assessment of whether the product achieved its strategic objective (Åstebro and Michela, 2005; Cooper and Kleinschmidt, 1995). While not all success outcomes are positively correlated and new products may succeed on one dimension but fail to meet their goals on another (Cooper, 1984), low-selling new products are unlikely to survive in the long term (Hart, 1993; Lomax et al., 1997).

This research investigates the role of the two most commonly used and easily accessible brand performance measures in predicting LE success and failure. These measures are penetration (i.e. number of households that purchase in a given period) and repeat-buyer rate (i.e. number of households who repeated purchase in a subsequent period). Our study extends past research that examined penetration and loyalty metrics for brand extensions (e.g. Singh et al., 2012), and another study identifying which types of customers during launch can help predict likely new product survivors and failures (Anderson et al., 2015).

Cumulative evidence about how brands perform in developed and developing markets consistently shows that changes in penetration contributes more to manufacturer and private label brand market share growth than changes in loyalty (e.g. Romaniuk et al., 2014, 2018). However, some new products might require a different strategy to survive. It is possible that new products need to stimulate greater changes in the repeat-buyer rate in the early stages of launch, to establish credibility and justify ongoing marketer and retailer support for survival.

As the studies outlined above have not specifically examined LEs, we investigate whether the metrics of repeat-buyer rate and penetration are related to long-term survival, and how early is it possible to detect a likely success. If one of these two commonly used performance measures plays a more significant role in new product survival, it would have major implications for strategy and resource allocation.

As an example, if the repeat-buyer rate is more important than the percentage trialing, this would suggest marketers should prioritize investing in superior product quality to ensure the first purchase is sufficiently satisfying to encourage subsequent repurchasing, rather than merely offering an adequate, safe product. This would increase the importance of past studies investigating the role of product quality and product innovativeness in new product success.

However, if changes in penetration are shown to be more important than changes in loyalty, this would suggest that marketers should invest more in activities to reach a wider category buyer base to attract as many new buyers as possible in each period. For example, this would include prioritizing marketing activities such as broader distribution and wide-reaching advertising. This recommendation would align with the cumulative knowledge supporting the market-based assets theory of brand competition (Sharp et al., 2024).

If both are equally important and cannot be used to detect the likelihood of survival for a new product introduction, this would also be an important finding. This finding would suggest that newly introduced consumer products require a unique dashboard of metrics during their launch phase, compared to the approach and metrics marketers use to measure the performance of established products and brands. This is investigated in the current study.

This study investigates the following research questions using consumer buying data:

This research extends existing knowledge about the role of penetration and loyalty in brand performance to include the role of these metrics on the likelihood of LE survival. The findings of this study are valuable to marketers seeking to identify LE survivors early on.

We investigate consumer purchasing behavior from 2004 to 2016 across five consumer goods categories: cookies, ground and whole bean coffee, ready-to-eat cereal, toothpaste and spray air fresheners. The sample spans food, beverage, personal care and home care categories. These categories are typically examined in industry innovation research (e.g. Circana, 2024). The categories vary in terms of purchase cycles and the level of new product activity, which is critical to our inbuilt replication approach for testing the generalizability of our results (Uncles and Kwok, 2013). We use the NielsenIQ Consumer Panel Data from the Kilts Center for Marketing (2025) because it includes extensive purchasing information for over a million consumer products. This data has been used in recent extension research (e.g. Koschmann and Sheth, 2018; Victory et al., 2024).

All analyses were conducted on a regional basis, as true national launches are rare in the United States (Hoskins et al., 2020), making penetration on a regional level more relevant. Investigating success at a regional level for new product introductions has been considered as early as Fourt and Woodlock (1960). We investigate LEs in the four most populated market areas in the United States: New York, NY; Chicago, IL; Los Angeles, CA; and Houston, TX. Although analyses were conducted by region, very little variation was observed across the four regional markets. The aggregate results are presented and discussed below.

As performance measures and success expectations vary by product type, this research focuses on the launch year performance of LEs. LEs are individual products that supplement an established brand’s product portfolio (Reddy et al., 1994; Sezen et al., 2024; Tauber, 1981) and include new flavors, packaging options or sizes (Rahman and Areni, 2014). LEs are vital to examine because most new consumer products are launched under an established brand name (Kovalenko et al., 2022; Nielsen, 2019).

Unfortunately, the secondary consumer purchasing data does not provide an explicit indicator to identify LEs from other products. In this study, LEs were identified in the consumer panel data as products (Universal Product Codes) from an established brand within the category that reported zero sales for at least one year, but that then began recording sales at a later date (e.g. Hoskins and Griffin, 2019; Hoskins et al., 2020; Sinapuelas and Sisodiya, 2010; Victory et al., 2021, 2024). Using this method, a total of 36,994 LEs in the data were identified.

We define success as survival time and measure changes in performance using two metrics:

1. repeat-buyer rate; and

2. penetration.

The quarterly performance of the survivor LEs and earlier failed LEs is compared to distinguish if (any of) these metrics differ between the two groups, and the timing for when any differences become apparent over the first year. We focus on first year performance because the window to gain consumer acceptance (e.g. Hoek et al., 2003; Victory et al., 2024), and retailer delisting decisions (Davies, 1994), supposedly closes within the first 12 months after launch. The two metrics are used to detect changes in category buyer adoption and subsequent repeat purchasing.

Survival (dependent variable): There was no explicit indicator in the data set to denote whether a product had been discontinued. This approach to measure overall success deviates from most studies, which rely on practitioner memory and expertise to recall success. Survival time is the dependent variable used to measure success, as LEs cannot continue to meet their objectives if they are no longer sold. The method for identifying survival using secondary data is similar to that in past research (e.g. Asplund and Sandin, 1999; Goldenberg et al., 2001; Salnikova et al., 2019; Victory et al., 2021, 2024; Wilbur and Farris, 2014).

Total regional sales for each LE identified were monitored quarterly from launch up to until three years later. LEs that failed before completing their first year, or LEs that survived to their second year but failed before reaching their third year, were excluded in this analysis. This approach also excludes seasonal or limited time offer LEs that were not intended to be around in the market long term.

Failed LEs are defined as LEs still reporting sales in the panel data one year after launch, but not continuing to report sales beyond their second year after launch. Survivor LEs are defined as LEs still reporting sales in the panel data three years after their launch.  AppendixFigure A1 illustrates how failed and survivor LEs are identified. While many LEs survive their first year, only around 20% of LEs meet the “survivor” status criteria used in this study (Victory et al., 2024). The industry also acknowledges that a substantial decline in sales and new product support typically occurs between the first and second year after launch (Nielsen, 2018), supporting our survivor classification. These one and three year windows are also consistent with precedent in past research comparing customer attitudinal indicators of launch survival (e.g. Anderson et al., 2015). Under this definition, 5,294 failed and 7,195 survivor LEs were identified and analyzed in this study.

Repeat-buyer rate (independent variable): The repeat-buyer rate measure used in this research calculates the percentage of households in the current period who had purchased the product in any previous period (East and Hammond, 1996). Similar studies investigating new brand launches apply a comparable loyalty metric to evaluate performance (see, Hoek et al., 2003; Singh et al., 2012). Repeat is also a common measure used in practice to assess new product success (e.g. Circana, 2024). This metric quantifies the retention of LE buyers after launch and provides insight into how each LE retains buyers from previous periods during the launch year. This was calculated using purchasing information in the panel data.

Penetration (independent variable): Penetration is defined as the number of households who had purchased the LE in the current period divided by the total number of category buyers in that period (Ehrenberg et al., 2004; Farris et al., 2016). Penetration (sometimes called “trial” for new launches) is a widely used metric for measuring new product performance in both industry (e.g. Circana, 2024) and academic research (e.g. Golder and Tellis, 1997; Lee and O’Connor, 2003). Penetration is particularly important for LEs as they are typically designed to appeal to the mass market (Hultink et al., 1998, 2000). Penetration was not precalculated in the data and was therefore calculated for each LE using the consumer purchasing data.

An empirical-then-theoretical (“empirics first”) approach is adopted to examine the research problem (see, Ehrenberg, 1994; Golder et al., 2023; Sharp et al., 2017). This approach is chosen because we address a significant (high cost and common) marketing problem. The findings from this study are expected to have immediate practical relevance for marketers, and both the data and language used are relevant to industry practice (Golder et al., 2023).

This study identifies and compares LE survival in Many Sets of Data to validate the existence of any law-like empirical patterns (Bass, 1995; Ehrenberg, 1990, 1995) through the inbuilt replication (Uncles and Kwok, 2013). Although this research approach is not required to be proceeded by established theory (Bass, 1995; Ehrenberg, 1995), this paper extends existing knowledge about established brands (e.g. Romaniuk et al., 2014, 2018) and brand extensions (e.g. Singh et al., 2012) to a new product type: LEs.

The mean performance on each metric for the “average” survivor and failed LE are compared, followed by a comparison of the performance distributions for each group. The distributions are shown to ensure the potential variability in metrics for each LE are not disguised in the mean figure. Since the distribution results were similar across the categories, an illustrative example is provided. The simplicity of our primarily descriptive approach is intended to demonstrate whether complex analyses are necessary to detect likely “survivors” soon after launch. If these simple methods are sufficient, this has great value to marketers who invest in expensive research to guide new product support and deletion decisions. Finally, a binary logistic regression is used to model the relationship between survival and the metrics of penetration and repeat-buyer rate.

Market share plays a role in survival and retailer stocking decisions (Asplund and Sandin, 1999; Davies, 1994). We first examined market share, as both trial and loyalty underpin this metric. Our analysis shows that the “average” survivor LE had a 0.4% share (SD = 0.3) directly after launch. This share remained similar in each quarter during launch. The “average” failed LE had a similar market share in the same quarter after launch (M = 0.3, SD = 0.2), but this declined to 0.2% by the end of the year (SD = 0.1). The variation in market share underscores the value of investigating repeat-buyer rates and penetration metrics as early indicators of survival.

The role of repeat buying in LE survival was investigated to address RQ1. The results are shown in Table 1. Across the categories, 18% of buyers of the “average” survivor LE had purchased the LE in the launch quarter and repurchased it in the subsequent quarter (SD = 5). In other words, 82% of the buyers of the “average” survivor LE in the quarter after launch had not bought it in the inital launch quarter. This rate is similar to the one in five buyers of the “average” failed LE, who purchased the LE in the launch quarter and repurchased it in the first quarter after launch (SD = 9). The repeat-buyer rate for both the “average” survivor and failed LEs continue to marginally increase each quarter following the launch, with somewhat similar rates for the survivor (M = 30, SD = 8) and failed LEs (M = 25, SD = 6). The initial comparison of the mean repeat-buyer rates suggests that the percentage of repeat-buyers are similar for both the “average” survivor and failed LEs over the launch period.

The results for the “average” survivor and failed LE in Table 1 might be criticized for masking the variability in the repeat-buyer rate, as the single repeat-buyer rate figure does not capture the variation for each survivor and failed LE. An illustrative example of the repeat-buyer rate distribution is shown in Figure 1 to show the variability in the measure for each LE. Figure 1 shows that the repeat-buyer rate distribution for the survivor and failed LE was nearly identical in the quarter directly after launch, but the distribution shape continued to change over the launch year. By the end of the launch year, failed LEs peaked below the 10% repeat-buyer rate, while survivor LEs exhibited a somewhat more standard distribution (i.e. more survivor LEs had a repeat-buyer rate near the mean and median). These results suggest that retaining repeat customers plays a role in LE survival. The importance of repeat-buying, relative to gaining new customers, is compared next.

Next, we compare the mean penetration achieved by the “average” survivor and failed LE, as shown in Table 2. In the quarter after launch, the “average” survivor LE achieved a 0.8% penetration (SD = 0.3), which is in stark contrast to the 0.4% penetration of the “average” failed LE during the same period (SD = 0.2). Importantly, while the “average” survivor LE sustained or increased its penetration through to the end of the launch year (M = 0.8%, SD = 0.4), the “average” failed LE saw a steady decline in penetration (M = 0.3%, SD = 0.1). The clear contrast in the penetration between the “average” survivor and failed LEs over the launch year shows the importance of gaining and sustaining penetration early to secure long term LE survival. The variability in LE penetration was more pronounced than the repeat-buyer rate results.

The single penetration figure in each quarter for the “average” failed and survivor LE may not fully represent the typical penetration these LEs receive at launch. In other words, the “average” figure in Table 2 does not show the spread in penetration across the LEs. An example demonstrating the variation shown in mean penetration figures is shown in Figure 2. The distributions of the penetration achieved by survivor and failed LEs both exhibit a unimodal shape that has a positive skew (see Figure 2). While there is considerable spread in what survivor and failed LEs achieve, the majority of the LEs have a penetration on the lower end. Although the penetration distribution has a similar shape for both the survivor and failed LEs, the distributions across the launch year reveal that fewer survivor LEs have a trial rate below 0.5% compared to failed LEs. This difference in trial between the two groups becomes more apparent as the year progresses.

We then compare how many survivor and failed LEs achieved a penetration above or below the category mean (see Table 3). Nearly 90% of failed LEs had a penetration below the category average one year after launch (SD = 2), compared to 52% of survivor LEs in the same quarter. The percentage of failed LEs falling below the category mean steadily increased over the launch year (M = 78%, SD = 6). A chi-square test for independence comparing the percentage of survivor LEs and failed LEs with a penetration below the category mean in the final launch quarter was statistically significant (1, n = 8636) = 1011.37, p < 0.001.

Altogether, the results above address RQ2 and suggest it might be possible to detect survivor LEs as early as the first year after launch. Although not all survivor LEs achieve a high penetration exceeding the category mean, failed LEs overwhelmingly exhibit quarterly penetration levels below the mean. This suggests this simple descriptive tool may be effective in diagnosing likely failures, helping marketers to either implement remedial action to change the trajectory of an initially underperforming LE or make an earlier decision to discontinue it.

To further quantify the role of trial and loyalty in LE survival, we used binary logistic regression to predict the likelihood of survival. The model included the independent variables of LE penetration and repeat-buyer rate, as well as category type and parent brand market share at launch. Parent brand market share was included due to its established influence in driving new product trial (e.g. Sinapuelas et al., 2015). A covariate for food vs nonfood categories was also included to accommodate for differences in category survival rates.

The overall model containing all predictors was statistically significant, χ²(5, n = 12489) = 1499.84, p < 0.001, indicating that the model can distinguish between the LEs that survived and those that failed. The model explained between 11.3% (Cox & Snell R square) and 15.2% (Nagelkerke R square) of the variance in LE survival and correctly classified 67.3% of cases. All independent variables provided a statistically significant contribution. Overall, the logistic regression results support the descriptive findings, particularly the critical role of penetration in predicting survival. The model results are shown in Table 4.

The strongest predictor of LE survival was higher mean penetration during the total observation period, which recorded an odds ratio of 4.69. This indicates that for each additional 1-point increase in penetration, the odds of survival increased by 4.7, controlling for all other variables in the model. This finding suggests that LEs that successfully build penetration are significantly more likely to survive in the long term.

Mean repeat-buyer rate was also a statistically significant predictor, though its effect was much smaller with an odds ratio of 1.01. Although the repeat-buyer rate does have a role in survival, its effect is small compared to penetration and the contextual variables. This supports Table 1 which showed similar rates for the “average” survivor and failed LEs. The best identifier between survivor and failures is variation in trial. A significant positive correlation for penetration and repeat-buyer rate is shown, r = 0.19, n = 12489, p < 0. 001, with comparable or slightly stronger relationships observed within individual categories.

Two contextual covariates were included: category type (food vs nonfood) and parent brand market share (high, medium and small). Both had a statistically significant role. The odds of survival were 1.4 times higher for LEs in food categories. Furthermore, LEs from parent brands with higher market shares were 30% more likely to survive, controlling for all other factors in the model. LEs from smaller share parent brands were less likely to survive. Overall, the model confirms the primary role of penetration, and the secondary roles of repeat purchasing and category and brand context in LE survival.

LEs, the focus of this study, are often considered an easier pathway to survival than other types of new product launches. Although LEs continue to proliferate on retailers’ shelves, marketers still lack simple tools to detect likely winners and potential failures early in the launch period. This study investigated the launch year performance of longer surviving and earlier failed LEs.

This study demonstrates clear differences in the early performance of survivor and failed LEs. Early sales performance is known to play a role in deletion decisions (e.g. Asplund and Sandin, 1999; Davies, 1994) but this study goes further by comparing the relative importance of the number of buyers (penetration) vs how many of the buyers repurchase (repeat-buyer rate). We show that survivor LEs generally achieve higher penetration throughout the launch period than failed LEs. In contrast, failed LEs typically attract fewer new buyers after their launch and continue a downward trajectory. The repeat-buyer rate for the survivor LEs and failed LEs also hardly vary.

The lack of meaningful variation in repeat-buyer rates aligns with a key underlying assumption of the NBD-Dirichlet model: that purchase timing follows an as-if-random Poisson distribution (Goodhardt et al., 1984; Trinh et al., 2014). If individuals buy at irregular time periods, this suggests that the repeat-rates are dictated less by buyer desires and instead more by category buying circumstances. Marketers may influence if someone buys the brand, but not when a future purchase will occur. In this instance, repeat-buyer rate would be largely unaffected even for a successful launch. This underlying assumption supports the conclusion that repeat is a poor standalone diagnostic measure, as it is a function of category buying. Furthermore, since new launches are more likely to be bought by heavy category buyers (Tanusondjaja et al., 2016), who typically have wider repertoires, this would lead to longer interpurchase intervals.

Although survivor LEs exhibit slightly higher repeat-buyer rates, survivor and failed LEs show stability in loyalty after the first or second quarter from launch. Similar patterns have been observed in past research for new brands and brand extensions (e.g. Hoek et al., 2003; Singh et al., 2012), and this research extends those findings to LEs. Although this is the first time the limited variability in repeat-buyer rate has been shown for individual LEs, a similar result has previously been shown for brand extensions (Singh et al., 2012), and in a study comparing repeat rates for a new brand and a similarly-sized established brand (Wright and Peat, 2002). Although high repeat rates are often celebrated in industry reports (e.g. Circana, 2023), this current study confirms that early trial is indeed the biggest hurdle for securing LE survival in the long-term.

Our findings suggest that prioritizing activities aimed at increasing repeat purchases over trial is not a viable pathway to secure survival. The evidence from this study suggests LE survival is more dependent on attracting more new customers to trial, and less about improving the loyalty/repeat-rate of (previously new) customers over the launch year. This finding supports conclusions in past research about the primary importance of penetration in brand extension market share. This is best summarized in past research that argues that while higher loyalty ratios are important, they are “a necessary (but not a self-sufficient) condition for success” (Fourt and Woodlock, 1960, p. 32). In other words, focusing on building repeat alone is not enough to encourage survival.

Finally, since early trial rates can quite quickly distinguish survivor and failed LEs, this implies that brands “cannot sneak the product out quietly and hope it will build up market share over time” (Hoek et al., 2003, p. 64). Although brands should invest in product quality and customer experience, disproportionate investment in these areas to boost repeat is unlikely to markedly improve LE survival. This is particularly true if such investments in product quality come at the expense of efforts to build awareness and attract new customers. Although LEs launched by higher share parent brands are more likely to produce LEs that survive, likely due to the stronger market-based assets of the parent brand (see, Sharp et al., 2024), this does not eliminate the need for continued investment in trial-focused activities.

This study investigates the first-year launch performance of longer surviving and earlier failed LEs. This research responds to previous calls in the literature for the need to evaluate new product success using explicit performance outcomes (e.g. Åstebro and Michela, 2005; Victory et al., 2024). The findings of this study offer three key contributions to marketing practice, which can assist in setting strategy, guiding investment decisions and monitoring success.

First, our findings highlight that changes in penetration serve as an important early performance indicator for new launches. This suggests that strategies geared at promoting new launch success should prioritize the recruitment of new buyers. LEs that survive follow a similar trajectory for growth as for established brands. Despite the importance of product quality and customer experience to encourage trial and foster “normal” repeat purchasing, this study finds no compelling evidence that a LE’s initial loyalty predicts its future survival or failure in the long term.

Second, the findings suggest the importance of marketing activities designed to drive penetration for LEs. The role of penetration supports industry proponents for product trial (Circana, 2023) and emphasizes the need to prioritize high reach launch activities, such as securing wide distribution and communicating relevant messages in high reach media (see, Sharp et al., 2024). As an example, seeing a new product in-store or advertised on TV are among the most common sources for new product awareness (Nielsen, 2015). Past research has established the critical role of securing distribution in driving new product trial and sales (Wilbur and Farris, 2014; Sinapuelas et al., 2015). In addition to considering post-launch reach, marketers can also implement strategies that aim to prioritize reach as early as pre-launch, specifically about which brand should launch the LE. Our findings mirror past research that demonstrates the value using larger brands to launch new offerings (e.g. Hoskins and Griffin, 2019). Although LEs launched by higher share parent brands may have “head start,” this initial advantage cannot be relied upon and must be supported alongside subsequent high reaching activities post-launch.

Finally, this research finds that for longer surviving LEs, higher penetration occurs quite quickly and is evident as early as the first quarter after launch. The model results support the importance of penetration as a key predictor of survival. Altogether, this suggests that greater penetration is a clear signal of likely future survival, while lower penetration is a warning of possible failure. Marketers who track penetration early will be able to adjust their marketing and investment levels accordingly for the LE, to either capitalize on the early positive signs or to remediate the poor performance. Being able to pinpoint likely failures early provides the opportunity to consider deletion early, rather than continuing to invest in something that is not showing improvement. For successful LEs, marketers should not cease support once the early positive signals manifest. Instead, be prepared to support the launch for years to come, as new product trial continues beyond the first year (e.g. Circana, 2023).

This study provides new evidence about how to predict which new products are likely to survive soon after their launch. Although we investigated thousands of failed and survived LEs, this research is not without its limitations. The limitations and possible research directions are discussed next.

First, we encourage future research to investigate the relationship between early performance and LE survival in an even broader range of categories, including for services. There is far more knowledge about the success of physical new products in consumer goods categories, including durables (e.g. Talay et al., 2024), rather than introductions in service categories where product offerings are intangible and there is no transfer in ownership (Kotler et al., 2010; McDonald et al., 2001). The distinct characteristics of service products may change the survivor criteria found in this study.

The second opportunity to expand the applicability of the findings of this research is by extending the product scope. In this study, we examined LEs introduced under an established brand name, which might provide an initial “head start” in trial (e.g. Ambler and Styles, 1996; Grasby et al., 2022; Smith and Park, 1992). However, these new products are typically not very unique or different to other products available in the market (Griffin and Page, 1996). Future research could test the survivor criteria from this study to the adoption and survival of very unique products, as these may take longer to establish and gain acceptance among consumers.

Another avenue of interest is investigating the survivor benchmarks for new products introduced by retailers, because their likelihood of survival and their delisting or adoption criteria will likely be different. One study shows new private label products have a higher failure rate than new national brand products (e.g. Salnikova et al., 2020), suggesting retailers are more likely to delist and churn out their own products earlier. Few studies report new private label penetration and loyalty performance (e.g. Trinh et al., 2016) but there is evidence that established store brands exhibit higher loyalty than expected (Dawes, 2022), perhaps due to the restrictions on physical availability. These findings begin to suggest that repeat loyalty is a more important measure for the survival of new private label products. This presents another brand type for future studies to investigate.

The similarity between the patterns in LE survival (this study) and brand extension success observed in past research suggests that perhaps the challenges and pathway to success for new and existing products/brands are similar. Marketers should plan “for penetration and increment reach through the whole portfolio, balancing existing and new, and ensuring relevance in multiple consumers’ needs” (Formisano et al., 2020, p. 135). However, for new introductions, it would be useful to understand whether new buyers of new launches require additional incentives to buy. Furthermore, it would also be valuable to better understand the source of “where” these buyers come. There is a need to better untangle this in future research, particularly in light of recent calls for large brands to consider their role in stimulating category growth as a driver of brand sales (Tanusondjaja et al., 2022).

A brand’s top selling product typically attracts the vast majority of its buyers (Tanusondjaja et al., 2018), and new variants are more likely to appeal to existing buyers (Trinh et al., 2016). This makes considering cannibalization critical. Recent research suggests that reducing feature similarity between products can help to reduce cannibalization (Sezen et al., 2024) and less substitutable products are less likely to have duplicate buyers (Aurier and Mejía, 2021). While some types of new products may have a better chance of growing brand market share (e.g. Gielens, 2012), there remains limited understanding about which product types are more likely to draw sales from certain products in the portfolio. More research exploring cross-purchasing within a brand's portfolio is required to help brands navigate maximizing coverage while reducing cannibalization.

Researcher(s) own analyses calculated (or derived) based in part on data from Nielsen Consumer LLC and marketing databases provided through the NielsenIQ Datasets at the Kilts Center for Marketing Data Center at The University of Chicago Booth School of Business. The conclusions drawn from the NielsenIQ data are those of the researcher(s) and do not reflect the views of NielsenIQ. NielsenIQ is not responsible for, had no role in, and was not involved in analyzing and preparing the results reported herein.