Who is winning the market game? Market importance-performance analysis as a new approach for multi-product market research

In today’s competitive marketplace, where products compete for consumer attention across crowded physical stores and digital shelves, understanding what makes one offering stand out from another has never been more important. Consider how Apple’s iPhone competes with Samsung’s Galaxy, or how Instagram competes with TikTok for user engagement. The success of a product depends not just on having good features, but also on understanding precisely which features matter the most to consumers and delivering superior performance where it counts. This competitive positioning has been an important topic for business strategists and researchers alike, but the tools we use to analyze product competitive advantages remain limited.

In market research and business literature, importance performance analysis (IPA) has long served as the go-to method to understand competitive advantages at the product attribute level (Bacon, 2003; Feng et al., 2014; Martilla and James, 1977). The method’s intuitive four-quadrant matrix has made it a favorite among practitioners in all industries. However, while real markets feature dozens of competing products, traditional IPA can only perform a one-on-one analysis (Albayrak, 2015; Mohammed et al., 2014; Smoliński et al., 2024a). Imagine trying to understand the smartphone market by only comparing the iPhone to Samsung’s Galaxy, ignoring Google Pixel, OnePlus and dozens of other products. This binary limitation severely limits IPA’s effectiveness in market analysis.

This paper introduces market importance-performance analysis (MIPA), an improved analytical framework that attempts to enhance how we measure and visualize competitive advantages in multi-competitor markets. Unlike existing approaches that force artificial one-on-one comparisons, MIPA can simultaneously analyze unlimited competitors. This methodological advance is possible thanks to operationalization of competitive advantage on two related levels: market-level expectations (what all consumers generally value) and product-level differentiation (how individual products create unique value propositions or conform to market expectations). Previous IPA applications conflated these two levels, overlooking business strategy insights that MIPA provides. For an example of these two levels, all smartphone users might value battery life (market-level), but Apple might strategically emphasize ecosystem integration (product-level differentiation).

The MIPA framework can identify these individual strategies and answer how products achieve competitive advantage. This is achieved through identifying performance superiority (doing things better than the rest) and importance differentiation (making different things matter or finding unique user segments). We show that successful products often combine both strategies in ways that create a competitive advantage in the market. For example, the social media Signal not only built a better privacy messaging platform (performance advantage), but also made attributes like privacy important to their consumers (importance advantage).

The remainder of this paper is organized as follows. We first present the theoretical foundations and mathematical framework of MIPA, detailing how it extends traditional IPA. Next, we provide interpretative guidelines for MIPA’s visual outputs. We then demonstrate MIPA’s research application and validate it through a case study of competing social media platforms. Our analysis concludes with a discussion of the theoretical and managerial implications of our findings.

The overarching goal of this paper is to extend and improve IPA, providing both practitioners and scholars with a powerful new approach to business strategy research. We aim to establish MIPA as a reconceptualization of how we measure, visualize, and strategize around competitive advantages. We acknowledge that it is still a new conceptualization with considerable development ahead. We discuss some of the limitations and future research directions at the end of the paper. Nevertheless, we believe the MIPA framework can handle the complexities of modern multi-competitor markets and brings us closer to answering the long-standing question of competitive analysis: Who is winning the market game?

In the 1980s and 1990s, the issue of competitiveness became one of the core research topics in business and economics (South, 1981). Competitiveness is a fundamental concept in market analysis, and is studied at multiple levels, from national economies and industrial sectors to individual businesses and their products (Buckley et al., 1988). At each of these levels, products form the foundational layer, as they are the means through which businesses compete for customers’ resources in the marketplace (Blackwell et al., 2006; Schiffman et al., 2013). Today, competitiveness is regarded an essential component of market analysis, with widespread calls to enhance competitiveness at every level of the economy, from enterprises and sectors to products and services. Successful business activities therefore rely on detailed information about market participants, their offerings, pricing strategies, marketing approaches and patterns of organizational development.

Government policies, sector-wide initiatives, and corporate business strategies generally aim to improve the competitive position of products and services in the market, creating more competitive businesses, sectors, and ultimately strengthening national economies. This product-centric view of competition aligns with the influential words of F.A. von Hayek, who described the market as a type of game in which all economic entities participate, with varying degrees of success proportional to their skills (Hayek, 1941). According to von Hayek, changes in consumer tastes, along with production factor prices, are fundamental aspects of the market game. He rejected the assumption that all products are homogeneous and that consumers are indifferent to the source of supply. He emphasized that products are inherently heterogeneous and consumers care deeply about their different features and attributes. Consequently, competition can be best described as a process in which producers adapt their offerings to meet consumer needs through various means, resulting in certain goods gaining more recognition from consumers than others. This ability to understand and respond to consumer preferences leads to a competitive advantage. If a producer cannot gain the acceptance of buyers, they are eliminated from the market game.

The most effective way to gain a competitive advantage is through the creation of an attractive product. In the contemporary marketplace, characterized by increasing competition due to a growing number of companies and continuous innovations, a competitive advantage distinguishes a product as more appealing to customers than alternatives. According to Porter (1985a, 1985b), there are three strategic approaches to achieving competitive advantage: cost leadership (optimizing operational efficiency to offer competitive pricing), differentiation (enhancing product performance and unique features), and focus (applying cost leadership or differentiation strategies to specific market segments). From the customers’ perspective, competitive advantages can be achieved through two customer-focused dimensions: better performance (superior product attributes), or increased importance (making products essential to users by marketing key features that resonate with their needs) (Albayrak, 2015). Furthermore, performance and importance advantages align with two of Porter’s strategies: a performance advantage enables differentiation through superior product attributes, and an importance advantage supports a focus strategy by identifying which product attributes matter most to specific users.

There exist three main methods for identifying competitive advantages: cognitive categorization-based, organizational identity-based (Mohammed et al., 2014) and comparative methods (Huarng et al., 2018; Lo et al., 2020). The first approach, employing a cognitive categorization-based method, relies on the mental models managers form of their competitors, often using imagined profiles and evaluating firms based on similarities. Although these methods benefit from managers’ direct market experience, they can be highly subjective. The second approach, using organizational identity-based methods, is rooted in economic theory and offers more objective analysis by assessing competitors based on market commonality (shared market characteristics) and resource similarity (shared technologies or resources). For example, Chen’s framework (Chen, 1996) emphasizes these two types of industry information, noting that companies with similar resources and market roles often exhibit similar strategic strengths and vulnerabilities. However, as Klier (2009) observes, modern businesses frequently operate in multiple markets with diverse product portfolios, making company-level comparisons increasingly inadequate. Today, many corporations are conglomerates that offer various products across different sectors, which requires a more granular analytical approach.

Such limitations have led to the adoption of the third approach: comparative methods. This technique focuses on comparative positioning and benchmarking, usually at more granular levels, allowing for a more detailed analysis of specific products and their features (Krishnamoorthy and D’Lima, 2014). Commonly used comparative analyses include the importance-performance analysis (IPA) (Keyt et al., 1994; Hemmington et al., 2018; La Fata et al., 2019), competitive feature analysis (Chen and Uysal, 2002) and product benchmarking studies (Hong et al., 2012). Comparative methods have a significant advantage; for example, while organizational identity-based methods might identify Apple and Samsung as competitors at the organizational level, comparative analysis methods allow for identification of competition at more specific product or feature levels, such as camera quality in smartphones, battery life in laptops or user interface design in software (Hai, 2023).

Among comparative methods, the Importance-Performance Analysis (IPA) has reached widespread popularity due to its focus at the product-level (Martilla and James, 1977). The basic premise of IPA is to measure customer perceptions of the importance of certain attributes, such as product quality, customer service, or pricing, and compare them with how well the product performs in these areas. IPA is typically visualized on a two-dimensional grid, divided into four quadrants (Bacon, 2003). In the “Concentrate Here” quadrant, we find attributes that are important to customers but where the product is underperforming. These areas require immediate attention. The “Keep up the Good Work” quadrant includes attributes that are both important and well-executed, representing strengths to maintain. In the “Low Priority” quadrant, listed are attributes that are neither important to customers nor well-executed, and thus require minimal focus. Finally, the “Possible Overkill” quadrant contains attributes where performance is high but importance is low, meaning resources spent here could potentially be redirected.

Originally introduced by Martilla and James (1977), the importance performance analysis has moved beyond its traditional quadrant interpretation and has evolved into a comparative method through several key contributions. Lowenstein (1995) expanded IPA’s theoretical foundation by emphasizing customer perceptions and positioning product attributes as key determinants of market success. Bacon (2003) advanced the practical application of the method by developing sophisticated visualization techniques, particularly by refining the IPA matrix as a tool to identify product advantages and disadvantages. A significant methodological expansion came from Deng et al. (2008), who transformed IPA into a benchmarking tool capable of systematic competitor comparison, marking IPA’s first integration with broader comparative analysis methods. Albayrak (2015) and Albayrak et al. (2018) further elevated IPA’s application in competitiveness research by developing Importance-Performance Competitor Analysis (IPCA), although this advancement is limited to binary competitor comparisons. Most recently, Phadermrod et al. (2019) broadened the analytical scope of IPA by demonstrating its compatibility with SWOT analysis frameworks. Through these developments, IPA has achieved an analytical standard framework that facilitates the identification of competitive advantages and disadvantages by simultaneously examining customer expectations of product attributes and their perceptions of actual performance.

IPA has been widely applied in both product and service evaluation in various industries such as tourism, healthcare, education, and technology. For instance, Duke and Mount (1996) proposed a general framework for product evaluation using IPA, while recently Smoliński et al. (2024a) applied an IPA-like framework to assess technology products. In the service sector, Deng (2007) used IPA in tourism to assess key service attributes, while Yavas and Shemwell (2001) applied it in healthcare to assess patient satisfaction with hospital services.

The simplicity of IPA has contributed to its widespread adoption. However, this very simplicity has also drawn criticism from researchers who point out several limitations. A particularly significant limitation is that current applications of IPA as a comparative method are restricted to comparing one product with a single competitor, failing to capture the broader market context needed for competitive positioning analysis (Albayrak, 2015; Phadermrod et al., 2019). This restricted comparison limits the utility of IPA in real-world markets where multiple competitors interact simultaneously. Additional criticisms include a lack of mathematical rigor and the absence of standardized measurement approaches (Eskildsen and Kristensen, 2006; Azzopardia and Nash, 2013).

Various researchers have attempted to address these shortcomings through methodological improvements, such as introducing regression-derived importance scores (Abalo et al., 2007), developing refined measurement scales (Hinderks et al., 2020), creating extensions that allow integration with SWOT analysis (Ahmed, 2021; Phadermrod et al., 2019; Uzun et al., 2024) and experimenting with Artificial Intelligence supported data collection (Januszewicz et al., 2025; Wu et al., 2023). Although these incremental improvements to IPA have addressed specific limitations, the core challenge of simultaneously analyzing multiple competing products remains unsolved. Attempting to understand competitive positioning through isolated pairwise comparisons is like trying to understand a chess game by examining only the movements of an single piece. True competitive advantage emerges not just from one-on-one comparisons but from a product’s position within the entire market. We suggest a different analytical sequence, one that first establishes market-level expectations before examining individual products. By understanding what constitutes expected performance and importance across all competitors, we create a reference framework that gives the position of individual products meaning. Only after mapping these market trends and heterogeneity can we properly assess whether a product’s performance represents genuine competitive advantage, disadvantage or conformity to market standards. Below, we introduce market importance performance analysis (MIPA) which operationalizes this perspective and provides a framework that first quantifies market expectations and heterogeneity, and then positions individual products within the market.

To illustrate our new approach, we first discuss a hypothetical example. Let P represent a set of distinct products from the same market (e.g. smartphones or social media applications). For each product $p∈P$⁠, we can evaluate its performance based on relevant attributes (e.g. ease of use or battery life). This assessment can be conducted through questionnaires (Bacon, 2003), automated text analysis (Smoliński et al., 2024b; Wu et al., 2023), or any other method that yields a user’s sentiment score for each attribute (Kangale et al., 2016; Kim et al., 2025). We then determine the importance of each attribute, either by directly asking users how important that attribute is for them, or analytically, by regressing the performance score on overall satisfaction measurements (e.g. product rating or use intention) (Dolinsky, 1991; Bacon, 2003). This process results in a collection of performance and importance scores for competing products across relevant attributes that likely influence a product’s success in the market. We assume that the resulting data exhibit significant heterogeneity when comparing the products using attributes’ performance and importance scores.

Performance heterogeneity reflects the varying degrees to which competing products within a market exceed or underperform on specific attributes versus each other and the market as a whole (competitive advantage versus competitive disadvantage). Such heterogeneity can be quantified by measuring deviations from the market’s expected value, typically calculated using the median or grand mean.

For products competing in the same market for similar users, a grand mean can be considered the market’s expected performance on that attribute. Deviations from this grand mean quantify market variation, with the standard deviation serving as a measure of market heterogeneity. If market heterogeneity is significant, we can investigate products with above or below average performance. We introduce a new measure called the market standardized performance score (MSP), calculated as follows:

This equation quantifies the p product’s performance (⁠$performancepj$⁠) deviation from the market’s expected performance (⁠$Performancejmarket$⁠) for a specific attribute j, scaled by the market’s standard deviation (⁠$σjPerformancemarket$⁠) for that attribute. In other words, MSP quantifies how many standard deviations a product’s performance deviates from the market’s expected performance (grand mean) for a specific attribute.

When $MSPj$ equals 0, the performance of the product in attribute j is exactly equal to the mean of the market. MSP values above 1 or below −1 signify that the product’s performance is more than one market deviation above or below the grand mean respectively. These deviations from the market expectation may indicate a competitive advantage or disadvantage, particularly if the attribute in question is highly important to product users.

However, the interpretation of MSP values should consider both the significance and magnitude of market heterogeneity. In markets with low heterogeneity, even small deviations from the grand mean can result in a product being labeled as having a competitive advantage or disadvantage. Conversely, markets with high heterogeneity require larger deviations from expected performance to achieve the same distinction. Nevertheless, even small deviations in a market with low heterogeneity can contribute significantly to a product’s advantage/disadvantage if the attribute is important for market success. Therefore, the interpretation of MSP as a competitiveness metric depends on:

• the overall importance of the attribute in the market;

• the significance and magnitude of market heterogeneity in performance; and

• MSP statistics for competing products.

Figure 1 presents the hypothetical distribution of MSP scores for four products. Products B and C have MSP values within one market deviation of the market grand mean, indicating that their performance aligns with market expectations (ranging from −1 to +1 market deviation). Product A has an MSP below the −1 market deviation threshold, suggesting that users perceive its performance for the attribute as inferior compared to other market offerings. If this attribute holds significant importance in the market and for the product’s user base, then the low MSP may signal a need to improve product performance. Product D, on the other hand, performs above the market expectation and has a competitive advantage.

Importance heterogeneity refers to variation in how users value specific attributes or features across different products within the same market. It is measured through differences in importance scores between products for a given attribute. Unlike performance heterogeneity, which primarily reflects underlying performance advantages or disadvantages, importance heterogeneity indicates divergent market trends and the potential formation of subcategories within a market.

When a product exhibits uniquely high importance scores in certain attributes compared to the rest of the market, it suggests an attempt to differentiate itself by prioritizing specific attributes. To identify such products, we can use a metric similar to that used for performance heterogeneity: the Market Standardized Importance (MSI) score:

This equation quantifies the p product’s importance (⁠$importancepj$⁠) deviation from the market’s expected importance (⁠$Importancejmarket$⁠) for a specific attribute j, scaled by the market’s standard deviation (⁠$σjImportancemarket$⁠) for that attribute. In other words, MSI quantifies how many standard deviations a product’s importance score for a specific attribute deviates from the market’s expected importance.

Once the market standardized importance (MSI) and performance (MSP) scores are calculated, we can analyze market positions by examining their joint distribution. Table 1 contains MSI and MSP scores for 10 hypothetical products on a relevant market attribute, visualized in Figure 2. The MIPA framework therefore offers two possible perspectives for understanding competitive dynamics, as illustrated in Figure 2. Panel (a) presents an attribute-centric perspective, with each node representing a competing product measured on a single attribute. It focuses on comparative positions and competitive clusters, answering the “who is winning the market game?” question by showing which products hold advantages in this specific attribute and identifying their direct competitors. Panel (b) adopts a product-centric perspective familiar to IPA practitioners, where each node represents an attribute for a single product, but now using MIPA’s standardized metrics (MSI and MSP) instead of scores for one product. This view focuses on where a product’s strengths and weaknesses lie across multiple attributes.

The MIPA matrix is divided into distinct zones through a combination of visual cues. A white circle at the origin (MSI = 0, MSP = 0) marks the market expectation point. It represents the average performance and importance levels across all products in the market. The matrix is segmented by dotted lines at MSI and MSP values of ±1. A central gray square marks the area where both MSI and MSP values fall between −1 and 1. It encompasses products that conform to market trends and expectations. Outside this gray area, quadrants are color coded to signal competitive positions: green zones indicate competitive advantage, red zones represent competitive disadvantage and yellow zones suggest ambiguous positions.

In the market perspective (Panel a), products with MSI and MSP values between −1 and 1 (e.g. products 5, 8 and 10) represent “Default Market Products.” For a hypothetical attribute like display quality, the gray zone would be mid-range smartphones which typically deliver expected display performance that users value at market-average levels. These products appear in the central gray area around the market expectation center in Figure 2.

Products outside of this gray area reveal distinct competitive positions. A “Performance Maximizer” like Product 7 (MSP > 1) appears in the green-shaded region on the right side of the matrix and might be analogous to a smartphone with an exceptional industry-leading OLED display. These products deliver superior performance on an attribute valued by the market. Conversely, Product 3’s underperformance (MSP < −1) places it in the red-shaded zone on the left, representing a phone with a subpar display quality requiring improvement.

In the green competitive advantage zone there is also a special category of Successful differentiation that appears positioned in the upper-right green quadrant (Product 1; MSP > 1, MSI > 1) – this might be an example of gaming-focused phones with high-refresh-rate displays that both perform well and are highly important to their gaming-oriented user base. Conversely, Product 2 illustrates failed differentiation, analogous to early foldable phones which increased the importance of display quality for its users (who paid premium prices) but suffered from significant display issues like creasing and fragility.

However, some MSI and MSP score combinations can be ambiguous in terms of market position. Tradeoff differentiation emerges in Products 6 and 9 (MSI < −1), both appearing in the lower yellow-shaded regions. Product 6’s low performance matches its low importance, like a budget phone where the display might have lower resolution, but users also expect less due to the lower price category. This is an ambiguous position because we deliver worse performance on a less relevant attribute for our users; however, the attribute still might be significant and impact acceptance. This ambiguity is even stronger for Product 9 with high performance despite low importance. This category mirrors IPA’s “Possible Overkill” quadrant, where attributes receive exceptional performance despite being less important to the product’s users. Tradeoff differentiators (MSI < −1) are particularly ambiguous – they might indicate either a viable niche where users genuinely expect less from products or a disadvantage if the attribute remains highly important despite its lower relative importance compared to market averages. Another ambiguous position occurs when MSI > 1 but MSP is average (−1 to 1), as with Product 4, where the product has successfully increased user expectations while having average performance. Product 4 therefore represents incomplete differentiation, akin to a phone that markets itself as having a ”cinema-quality display” thus heightening display expectations, but offers only standard display found in other phones.

Panel (b) illustrates the product-centric perspective, where we can examine a single product’s positioning across all attributes. This view reveals where the product’s strengths and weaknesses lie. For instance, Product 1 (which has an advantage in display quality with high performance and importance), excels in Attribute 1 (display), but the rest of its attributes fall within market expectations. Additionally, it faces a competitive disadvantage for Attribute 2. This attribute represents clear underperformance – perhaps battery life, where the high-performance display’s power consumption creates a tradeoff. This product-centric view allows managers to identify which attributes require attention and which represent competitive advantages.

Table 2 provides a quick reference for interpreting different combinations of MSI and MSP scores. We have assigned intuitive names to products with various combinations of MSI and MSP scores, such as “Performance Maximizer” or “Failed Differentiator.” For each combination, we also indicate the corresponding market position, specifying whether the product likely has a competitive advantage (green zones), competitive disadvantage (red zones), conformity to market expectations (gray zone), or if the position is ambiguous (yellow zones). In ambiguous cases, further analysis and comparison across other relevant attributes are needed to determine the product’s overall market position.

To demonstrate the power of Market Importance Performance Analysis’s framework, we present a case study of social media communication platforms. We selected this market for several compelling reasons. First, social media platforms represent a contained, easily analyzable market with clear competition. Second, these platforms are widely used and familiar to consumers, providing a rich source of user perceptions and experiences. Third, social media platforms actively compete through both performance improvements and feature differentiation, making them ideal for studying competitive advantages. Finally, the market exhibits significant heterogeneity in user preferences and platform positioning, allowing us to evaluate the ability of our analytical framework to capture competitive advantages and disadvantages (Zhang and Sarvary, 2011).

To define relevant attributes for comparing social media platforms, we draw upon the Technology Acceptance Model (TAM). TAM has become a standard approach for analyzing technology adoption, validated across numerous contexts including social media platforms (Kwon et al., 2014; Makki et al., 2018; Smoliński et al., 2023). TAM uses Behavioral Intention (BI) – users’ intention to use a technology – as its primary outcome variable. In social media markets, where platforms compete primarily for user attention and engagement, BI can serve as an appropriate approximation of market success.

Based on a review of the literature, we selected six attributes consistently identified as significant predictors of platform adoption: Perceived Usefulness (PU), Perceived Ease of Use (EU), Perceived Enjoyment (PE), Social Influence (SI), Technology Attachment (TA) and Security and Privacy (SP). These attributes collectively capture the multidimensional nature of social media platform acceptance, spanning functional, hedonic, social and security considerations. Their extensive validation in multiple studies as determinants of behavioral intention and subsequent market success establishes them as reliable market attributes for our analysis (Al-Qaysi et al., 2020; Smoliński et al., 2023). Detailed definitions and literature foundations for each attribute are provided in Table 3.

Our selection of both social media market and TAM variables represents just one possible application of the MIPA framework. The social media market could be analyzed using differing sets of attributes depending on research objectives. Some analysts might focus on technical features, others on content characteristics, and yet others on monetization potential. The MIPA framework can be applied to any market with competing products, using any set of relevant attributes for comparison. However, two key considerations should guide market and attribute selection.

First, market definition must be comprehensive enough to establish meaningful market reference points for performance and importance scores. When relevant products are omitted, our reference estimates become biased as we fail to account for significant competitors in the market. This means our assessment of competitive advantages or disadvantages becomes limited, only showing how products compare to a subset of the selected competing products. That is a reason why establishing a thorough market definition and capturing as many relevant products as possible is crucial for meaningful analysis.

Second, the selected attributes require careful consideration. They must be relevant across all competing products to enable consistent comparison and must be measurable through user perceptions and sentiment. These attributes should significantly influence user satisfaction and product success. Furthermore, they should be capable of revealing meaningful differentiation between products, allowing the analysis to capture distinct competitive positions and strategies.

Having defined the social media platform market and the relevant attributes for market success, we now turn to developing quantitative models for the Market Performance Importance Analysis. Our approach involves defining two interconnected models: a market performance model and a market importance model. These models estimate the expected performance and importance of attributes across the market and for individual social media platforms, accounting for market heterogeneity.

The performance model is specified as follows:

where:

• $Pijk$ is the performance score for attribute j, given by user k for platform i;

• $βj$ is the fixed effect for attribute j, representing the market-level expected performance (⁠$Performancejmarket$ from the MSP equation);

• $uij$ is the random effect for platform i on attribute j, corresponding to $performanceij−Performancejmarket$ from the MSP equation;

• $vk$ is the random effect for user k, controlling for individual differences in rating behavior; and

• i indexes products, j indexes attributes (PU, SI, PE, PT, EU, SA), and k indexes subjects.

The market performance heterogeneity for attribute j (⁠$σjPerformancemarket$⁠) is quantified by calculating the standard deviation of the platform-attribute random effects $uij$⁠.

The importance model is specified as:

where:

• $BIik$ is the behavioral intention score given by user k for platform i;

• $β0$ is the model intercept;

• $ui0$ is the random intercept for platform i;

• $βj$ are the fixed effects representing market-level importance weights (⁠$Importancejmarket$ from the MSI equation);

• $Xjik$ represents the centered performance score for attribute j;

• $uij$ is the random slope for platform i on attribute j, corresponding to $importanceij−Importancejmarket$ from the MSI equation; and

• $vk$ is the random effect for user k.

The market importance heterogeneity for attribute j (⁠$σjImportancemarket$⁠) is quantified by calculating the standard deviation of the platform-attribute random slopes $uij$⁠.

Both models employ a skew normal distribution to accommodate asymmetric distributions commonly observed in user perception scores, where ratings tend to cluster toward the positive or negative ends of the scale (Smoliński et al., 2023). The complete specification of these models, including prior distributions and estimation details, is provided in the Supplementary Materials.

At the heart of our approach lies heterogeneity estimation, making both our models special cases of mixed effects models. The performance model is a mean difference model with random effects for platform-attribute combinations and the importance model is a regression model with random effects for the same combinations. Given this random effect structure, using analytical estimation (e.g. maximum likelihood estimation) would be cumbersome, so we opted for Bayesian estimation methods. These methods allow us to simply specify model equations and estimate parameters with their credible intervals using Markov Chain Monte Carlo techniques (McElreath, 2016).

Bayesian methods offer additional advantages when estimating market heterogeneity based on user perception data. They can handle three challenging characteristics of such data: complex nested hierarchies (where users are nested within their respective products), crossed random effects (where users can be associated with multiple products simultaneously), and unbalanced sample sizes (where the number of users varies substantially across products). Traditional frequentist methods are less efficient than Bayesian models when handling such complex, multi-level data structures (Certo et al., 2017, 2024).

Another advantage of Bayesian models is their ability to incorporate prior knowledge and expectations about attribute effects. In this paper, we used weakly informative priors for all parameters, allowing the model estimation to be driven primarily by the observed data. Future research could specify more informative priors when domain expertise or previous findings warrant stronger assumptions about parameter values. For brevity, we omit the detailed description of the model estimation procedure and priors here; a detailed explanation of the priors and model specifications, along with our reasoning and R code, can be found in the Supplementary Materials.

We used a questionnaire developed by Smoliński et al. (2023) to assess technology acceptance of social media platforms. Previous research by Smoliński et al. (2023) demonstrated that this questionnaire has good validity and effectively measures six relevant market attributes: Perceived Usefulness (PU), Perceived Ease of Use (EU), Perceived Enjoyment (PE), Social Influence (SI), Technology Attachment (TA) and Security and Privacy (SP), as well as our defined approximation of market success: Behavioral Intention (BI). Each variable was assessed through four questions presented on a 7-point Likert scale. The scores on this scale represent users’ sentiment toward a studied attribute and Behavioral Intention (willingness to use the social media platform). Nine social media communication platforms were selected for analysis: Facebook Messenger, Instagram, Discord, Signal, Snapchat, Telegram, TikTok, Twitter (before its rebranding to X) and WhatsApp.

We collected a convenience sample of 614 social media users in Poland (January and February, 2023). The participants were mostly young adults who were students at universities in Poland and had volunteered to take part in our survey. The median age in our sample was 22 years old, with a mean of 23.54 years and a standard deviation of 5.75. In the social media applications market, some users might use more than one platform (e.g. using Facebook Messenger and WhatsApp). Facebook Messenger had the highest number of users among social media platforms included in our study (n = 602), followed by Instagram (n = 491) and WhatsApp (n = 359). Telegram and Signal had the lowest number of users (n = 83 and n = 54, respectively).

This sample may not be representative for all social media users. However, our objective is to demonstrate the MIPA approach and its analytical capabilities rather than to draw conclusions about the representative social media market. Our sample reflects the attribute importance and performance of social media platforms among young Polish users, which may vary from samples from different countries, cultures and age groups. This limitation does not necessarily undermine the study’s value. Practitioners often focus on results from specific user segments and groups. Understanding which factors determine product acceptance in these distinct groups is as important as understanding general market trends. Our sample approach may resemble how business practitioners would utilize this methodology. Given these considerations, we believe that the convenience sample of Polish students is adequate to illustrate and validate the MIPA framework. We later discuss alternative data sources and sampling strategies in the Discussion and Limitations sections.

Table 4 presents the results of the market performance model. We observe subtle variations between different platforms (term $uijP$⁠) around the market expectation (term $βjP$⁠, representing the grand mean). Ease of Use is the attribute with the highest expected market performance (term $βjP$⁠) with $(PerformanceEUmarket=5.31)$⁠, closely followed by Social Influence $(PerformanceSImarket=4.78)$⁠. These performance expectations indicate that users generally perceive most platforms as excelling in Ease of Use and Social Influence. This perception may be the result of platforms investing in these attributes, resulting in high performance ratings across the market. Conversely, Technology Attachment and Security and Privacy show the lowest expected performance scores. $(PerformanceTAmarket=3.89;PerformanceSPmarket=3.89)$⁠. These are the attributes that users find least satisfactory (performant) across the platforms. As a reminder, a comprehensive explanation of attributes is available in Table 3.

Heterogeneity in user-perceived market performance is significant for all attributes, justifying our application of Market Standardzied Performance scores to investigate the market in search of competitive advantages or disadvantages across all relevant attributes. We also observe significant subject heterogeneity, which means that individual users tend to rate platforms differently across attributes. For example, one user might have a tendency to rate platforms more positively on average than another user who has a tendency to rate platform performance generally less favorably. This observed subject heterogeneity supports our approach of using Bayesian mixed effect methods to control for these individual differences in ratings and thereby obtaining more reliable estimates of platform performance across the market.

The market importance model is presented in Table 5. Social Influence is the attribute with the highest expected importance $(ImportanceSImarket=0.28)$⁠, followed by Ease of Use $(ImportanceEUmarket=0.20)$⁠. These findings align with our results from the performance model, where Ease of Use and Social Influence demonstrated the highest expected performance. This suggests that on average, platforms are performing well in the attributes that are most important (on average) to users. It indicates a general market-wide understanding of user expectations and a corresponding allocation of resources to meet these needs. Conversely, Technology Attachment and Security and Privacy rank as the least important attributes on average $(ImportanceTAmarket=0.14;ImportanceSPmarket=0.13)$⁠. Again, such results correlate with our findings from the performance model where these attributes showed the lowest expected performance. We can therefore similarly infer that on average platforms tend to prioritize resources (and thusly performance) less on attributes that are, on average, less important to users.

The market heterogeneity section of Table 5 reveals that not all attributes have significant heterogeneity in platform importance. The two most important attributes, Social Influence and Ease of Use, do not present significant heterogeneity in importance scores across social media communicator platforms. Users tend to perceive the importance of these attributes similarly for each platform (homogeneity). It suggests that these are core attributes for the acceptance of all social media communicators. The lack of heterogeneity in these attributes might indicate that differentiation on these attributes is difficult, presumably due to their greater market significance (baseline importance) or the already high expected performance across platforms. However, significant heterogeneity is present in Perceived Usefulness, Perceived Enjoyment, and Security and Privacy. This heterogeneity suggests that individual platforms are attempting to differentiate themselves from the rest of the market on these attributes. We can further explore these differentiation strategies through the use of market standardized importance scores as part of our joint market importance performance analyses (MIPA).

Table 6 presents the Market Standardized Performance and Market Standardized Importance scores. These scores can be visualized from two perspectives. Figure 3 shows the perspective of the product, where each point represents an attribute for a single platform. This perspective is useful in identifying the strengths and weaknesses of the individual platform. Figure 4 displays the market perspective, where each point represents a platform for a single attribute. This perspective facilitates comparisons with competitors and focuses on competitive advantages.

Social Media Communicators like Discord, Instagram, Snapchat and Telegram show the expected market fit, with their MSP remaining within the expected market parity range (−1 to 1) regarding all attribute performance. Figure 3 illustrates how each platform’s performance profile spans across all attributes, and Figure 4 displays their competitive positions on individual attributes. Facebook Messenger stands out with significant performance advantages compared to market expectations in Perceived Usefulness (⁠$MSPPU=1.88$⁠), Social Influence (⁠$MSPSI=1.65$⁠), Ease of Use (⁠$MSPEU=1.59$⁠) and Technology Attachment (⁠$MSPTA=1.33$⁠). Given that Social Influence and Ease of Use are the most important attributes in the market importance model, Facebook Messenger likely has a significant competitive advantage, explaining its dominant market position (measured by users numbers and Behavioral Intention). Signal demonstrates above-expectation performance in Security & Privacy (⁠$MSPSP=1.68$⁠), compared to Telegram, which does not differentiate strongly on this attribute despite its reputation for privacy (⁠$MSPSP=1$⁠). TikTok shows a performance advantage in Perceived Enjoyment (⁠$MSPPE=1.20$⁠), aligning with its popularity as an entertainment-focused platform. Twitter and WhatsApp are the only platforms performing below market expectations, with Twitter underperforming in Ease of Use (⁠$MSPEU= −1.11$⁠) and Social Influence (⁠$MSPSI= −1.19$⁠) and WhatsApp in Perceived Enjoyment (⁠$MSPPE= −1.32$⁠). Twitter’s underperformance in Ease of Use and Social Influence, the most important market attributes, may indicate a serious competitive disadvantage.

Significant market heterogeneity is found only for Perceived Usefulness, Perceived Enjoyment, and Security & Privacy. This heterogeneity is indicated by MSI scores outside the −1 to 1 range for these attributes. Market Importance Scores from Table 6 indicate that the importance heterogeneity for each attribute is primarily driven by one app attempting to differentiate: Signal. For Signal users, Perceived Usefulness (⁠$MSIPU= −1.06$⁠) and Perceived Enjoyment (⁠$MSIPE= −1.62$⁠) are significantly less important than market expectations, although this does not mean that these attributes are unimportant; instead, Signal users find them less relevant compared to users of other platforms on the market. In contrast, Security and Privacy are more important for Signal users than for users of other platforms (⁠$MSISP=1.34$⁠), indicating that Signal has successfully differentiated itself by making this attribute a more important driver of acceptance compared to other platforms (market expectations).

Figure 4 presents Market Importance-Performance Matrices from the market perspective. Figure 3 complements this view by displaying the product perspective. Based on these visualizations and the combinations of MSP and MSI scores, we can identify market positions and conduct comparative analysis between platforms.

Facebook Messenger exemplifies a Performance Maximizer for four out of six market attributes: Perceived Usefulness (⁠$MSPPU=1.88$⁠), Social Influence (⁠$MSPSI=1.65$⁠), Technology Attachment (⁠$MSPTA=1.33$⁠), and Ease of Use (⁠$MSPEU=1.59$⁠). Two of these attributes (Social Influence and Ease of Use) are identified as the most important determinants of market success based on their importance weights from Market Importance Model (Table 5). These performance scores indicate a significant competitive advantage. This advantage becomes clear when comparing Facebook Messenger with a platform like Twitter, which underperforms Facebook Messenger by 2.84 and 2.70 market standard deviations on Social Influence and Ease of Use, respectively. The excellent performance of Facebook Messenger on these attributes likely stems from its large user base. Platforms with large user populations perform well on usefulness and social influence features, as these attributes are inherently tied to network size – even if competing platforms offer similar functionality.

TikTok adopts a distinct but equally viable performance maximization strategy, dominating the Perceived Enjoyment (⁠$MSPPE=1.20$⁠) and Technology Attachment (⁠$MSPTA=1.01$⁠) category. TikTok’s entertainment-focused approach creates a 0.47 standard deviation gap over Instagram and a 2.43 gap over WhatsApp on enjoyment. Other platforms with performance scores suggesting an enjoyment maximization strategy are Twitter (⁠$MSPPE=0.78$⁠) and Instagram (⁠$MSPPE=0.73$⁠). However, these platforms have not achieved the same competitive advantage as TikTok. This might change if they commit to improving features that maximize user satisfaction with this attribute.

In conclusion, both Facebook Messenger and TikTok can be classified as Performance Maximizers, as their competitive advantages come from superior performance on relevant attributes. Facebook Messenger may have a slightly higher competitive advantage in the overall user base and willingness to use the platform (Behavioral Intention) due to its focus on attributes with higher market importance. However, the overarching market strategies of these social media communicators are similar, leading to competitive advantages in their respective focus areas.

These platforms achieve market success by conforming to market trends, rather than differentiating themselves on performance (e.g. TikTok) or importance (e.g. Signal). Instagram emerges as the strongest default market product with consistently positive scores across five of six attributes (0.41–0.73 range). Discord similarly achieves near-optimal balance across all attributes for its target audience: moderate positive performance on Usefulness (⁠$MSPPU=0.38$⁠), Social Influence (⁠$MSPSI=0.35$⁠) and Perceived Enjoyment (⁠$MSPPE=0.36$⁠) and expected performance on Ease of Use (⁠$MSPEU= −0.13$⁠) and Security & Privacy (⁠$MSPSP=0.21$⁠). In contrast, Snapchat’s slightly negative scores on Usefulness (⁠$MSPPU= −0.56$⁠) and Ease of Use (⁠$MSPEU= −0.53$⁠) suggest a strategic drift from its original messaging innovation to now average performance across all features.

Telegram performs broadly within market expectations, but with slightly lower performance scores than Instagram and Discord across most attributes. However, Telegram demonstrates strong performance on Security & Privacy (⁠$MSPSP=1.00$⁠), positioning it very close to successful performance maximization on this attribute and potentially qualifying as a Performance Maximizer. This focus on privacy and security represents a strategic differentiation attempt. Nevertheless, Telegram still falls behind Signal, which maintains a 0.68 standard deviation advantage over Telegram (Signal: $MSPSP=1.68$⁠) and represents the clear leader in security and privacy features.

Signal presents this market’s only case of successful attribute differentiation. Signal achieves competitive advantage by simultaneously increasing Security & Privacy importance for its users (⁠$MSISP=1.34$⁠) and delivering superior performance on this attribute (⁠$MSPSP=1.68$⁠). However, Signal’s differentiation strategy involves tradeoffs, with decreased importance of Perceived Enjoyment (⁠$MSIPE= −1.62$⁠) and Perceived Usefulness (⁠$MSIPU= −1.06$⁠) compared to market expectations. This pattern indicates that Signal’s users demonstrate willingness to accept tradeoffs between enjoyment and usefulness features in exchange for improved privacy and security. Signal’s success validates the hypothesis that privacy-conscious users represent a distinct market segment willing to sacrifice convenience for security, a finding with potentially significant implications for discussions around data privacy.

Twitter finds itself in the most disadvantageous market position. It faces performance disadvantages in Social Influence (⁠$MSPSI= −1.19$⁠) and Ease of Use (⁠$MSPEU= −1.11$⁠). As we saw in the case of Facebook Messenger, these attributes are the main drivers of market success and user satisfaction. Low scores on these attributes suggest fundamental platform issues rather than strategic positioning. Twitter’s Social Influence disadvantage is particularly large. It trails Facebook Messenger by 2.84 standard deviations, Instagram by 1.68, and WhatsApp by 1.45, indicating potential for rapid user exodus if alternatives emerge.

WhatsApp faces more nuanced challenges. Its Perceived Enjoyment disadvantage (⁠$MSPPE= −1.23$⁠) creates a 2.43 standard deviation gap behind market leader TikTok, but it maintains the expected performance on Social Influence (⁠$MSPSI=0.26$⁠) and Ease of Use (⁠$MSPEU=0.37$⁠). This pattern suggests WhatsApp could improve its competitive position through entertainment feature integration – perhaps explaining Meta’s recent push toward WhatsApp Stories and Community features.

We have shown that MIPA can be successfully used to simultaneously analyze a number of competing products in the same market. Through comparative analysis, it can identify market positions and infer competitive advantages and disadvantages. The core innovation is that MIPA expands IPA into a multilevel analysis tool. We first identify market expectations, determine how certain attributes contribute to average user satisfaction, and assess how products perform on average for each attribute. Once we calculate market importance and performance expectations, we can compare individual products to these expectations. This allows us to identify which products overperform (Performance Maximizers), which underperform (Underperformers), and which attempt to differentiate themselves by emphasizing certain attributes (differentiators).

The MIPA framework allows for the visualization of market positions from two perspectives. The attribute, market-centric perspective: focusing on specific attributes (e.g. security and privacy features in social media platforms), enabling comparison between products to identify high performers, underperformers, and closest competitors. The product-centric perspective: zooming in on individual products to identify their strengths and weaknesses.

However, as noted in the introduction, MIPA is a new approach that still requires development. Although MIPA is a much more powerful approach to market research than IPA or similar methods like SWOT, this power comes at the price of complexity. It requires data on user perceptions and satisfaction from multiple products. Markets can be extensively large, with competing products numbering in the hundreds. In contrast, collecting data about one or two products, as traditionally done for IPA approaches, is relatively easy to implement. Market research of this type often involves asking existing users or beta testers about their product perceptions on relevant attributes. The researchers then create IPA matrices and perform analysis. However, such analysis does not tell us anything about competitors or the market position of products. It remains oblivious to market expectations, which must be factored in through cognitive categorization-based approaches. And, as discussed in the theoretical perspective section, these approaches can be flawed.

Implementing full-scale MIPA with traditional survey or focus group approaches would be cumbersome and costly. MIPA necessitates a better way to collect data on user sentiment and satisfaction. The answer comes in the form of online data collection. More concretely, the internet provides a large source of user reviews and ratings that can be incorporated into MIPA. Significant groundwork has already integrated IPA-like analysis with online data sources. Recently, Hu et al. (2020) proposed a sentiment-based scoring approach that extracts product performance scores from the online review text based on the frequency of attribute keywords mentioned. Bi et al. (2019), Albayrak et al. (2021), Joung and Kim (2021), and Zhang et al. (2021) proposed similar natural language processing (NLP) approaches to Importance Performance Analysis. Wu et al. (2023) addressed critiques about review credibility by incorporating measures of review reliability when calculating importance and performance scores.

Recent developments include work by Smoliński et al. (2024a, 2024b) and Januszewicz et al. (2025), who introduced Large Language Model (LLM) Annotation Systems to score online reviews. These systems combine local LLMs with scoring instructions to analyze text and extract sentiment scores on easily definable attributes. These methods require only attribute definitions and scoring scales. The LLM automatically scores reviews on these attributes using the provided scale. These advances in Big Data analysis and Artificial Intelligence have largely resolved concerns about MIPA’s feasibility. Collecting data on user sentiment and product satisfaction now requires only a simple scraping agent and sentiment-based or large language model-based annotation systems. Once annotation scores are calculated, the MIPA model can be estimated.

Estimating MIPA as introduced in this paper involves Bayesian models. We implemented a Bayesian approach because it handles complex data structures well. It requires only model equations and reasonable priors; Bayesian estimation methods such as MCMC handle the rest. However, this can be a barrier for practitioners trained in frequentist statistics and data analysis approaches. The next step in MIPA development might be testing more user-friendly estimation approaches resembling IPA estimation with simple averages from sentiment scores and aggregated regression averages instead of mixed-effect models. Future simulation studies should determine whether these suboptimal estimation approaches adequately approximate market-level and product-level effects.

Once the model is estimated, the MIPA metrics are straightforward to interpret. Practitioners and researchers can choose between MIPA visualizations based on their goals: the market-centric attribute-level perspective or the product-centric perspective. Market researchers will find the market-attribute centric perspective more useful, as it first reveals market expectation estimates, then product positions and clusters. Researchers can identify the most relevant market attributes by analyzing the market expectations weights from the importance model. They can then judge the performance of individual products on these important attributes to identify competitive advantages and disadvantages.

Business practitioners may find that the product-centric approach offers more targeted insights. From a product design perspective, identifying product attributes that are weak (below market expectations) or strong (above market expectations) provides useful guidance for product testing and improvements. Additionally, identifying attributes where products show differentiation (e.g. greater importance expectations than the market) tells us whether products successfully target their intended niche user groups.

Other analysis modes are also possible, and MIPA permits many extensions. Business practitioners often focus on product evolution and longitudinal market trends. MIPA can be easily adapted to longitudinal approaches by extending MIPA equations to include time series trends. Researchers can collect data on different generations of products from online reviews, annotate it with LLM models, and estimate longitudinal MIPA using Bayesian mixed-effect models. For example, our case study provides a static comparison of social media platforms. However, social media platforms constantly evolve through feature updates, interface changes, and rebrandings (e.g. Twitter changing to X). Static analysis identifies market positions at one point in time. However, longitudinal MIPA could track how these positions evolve. For instance, WhatsApp’s communities feature might have improved its Perceived Enjoyment performance from below to above market expectations, and Twitter’s rebranding to X could have altered its market differentiation strategy. Longitudinal MIPA would identify these market trends. By taking multiple snapshots over time, researchers could map the dynamic trajectories of platform development and market repositioning. Such analyses are possible and future studies should prove their viability.

Competitiveness remains an integral part of market analysis, underpinning von Hayek’s interpretation of markets as a game where success depends on understanding and responding to consumer needs. In modern markets, gaining competitive advantage increasingly relies on proper product creation (differentiation strategy) and attribute differentiation (focus strategy). Various methods exist to identify competitive advantages in products, yet they all suffer from some form of imperfection, and there remains a need to refine them to better serve researchers and practitioners. Our study focused on one such comparative method, Importance Performance Analysis (IPA), which relies on measuring user perceptions of product attributes’ importance and performance. Although IPA has proven valuable for comparing products with their competitors, its traditional application suffers from significant limitations, particularly in analyzing markets with multiple competing products.

To address these limitations, we developed Market Importance Performance Analysis (MIPA), a novel analytical framework that extends traditional IPA to create a proper tool for competitive advantage and market positioning analysis. The major contribution of MIPA is its departure from IPA’s focus on single-product or binary product comparisons. Our approach substantially redefines IPA to enable comparison of multiple products simultaneously. We introduce standardized metrics – Market Standardized Performance (MSP) and Market Standardized Importance (MSI) scores – that enable consistent comparison of products across studied attributes in reference to market-level trends and heterogeneity. The interpretations of our metrics remain consistent regardless of scale, data measurement, defined attributes or products under investigation. These metrics provide clear mathematical interpretations of products’ competitive advantages and differentiation strategies. Additionally, they allow for simple and intuitive visualizations through MIPA matrices.

We validated our framework through a case study analysis of social media communication platforms and demonstrated MIPA’s ability to identify distinct competitive positions and strategies. Our analysis revealed four key market positions: Performance Maximizers (Facebook Messenger and TikTok), which excel through better-than-expected attribute performance; Default Market Products (Instagram and Discord), which meet market expectations without significant differentiation; Successful Differentiators (Signal), which achieve advantage through attribute differentiation; and Underperformers (Twitter and WhatsApp), which fail to meet market expectations on key attributes.

The supplementary material for this article can be found online at https://osf.io/zp2cx