Decoding a Market Share Anomaly in Rural Telecommunications: Perception, Behavior, and Execution Drivers in Rural Banten, Indonesia

1. Introduction

Mobile connectivity has shifted from a convenience to a necessity in developing economies, where it underpins access to employment, banking, healthcare, and social participation. Yet a third of the global population remains offline, and the unconnected are disproportionately low-income and rural (Houngbonon, Ivaldi, Palikot, & Strusani, 2025). In Indonesia, an archipelago of roughly 17,000 islands and more than 270 million people, mobile networks are central to digital inclusion and to the equitable distribution of the gains from a rapidly expanding digital economy (Putri, Aziz, Simon, Aryatama, & Oktapiani, 2025). Mobile operators are therefore not only commercial actors but instruments of socioeconomic development.

Telkomsel has dominated Indonesia's mobile sector since 1995, serving more than 158 million customers and holding a 48.6% national market share in 2025. National leadership, however, does not guarantee local dominance. Regional outcomes are shaped by disposable income, cultural norms, retail accessibility, and perceived value, which can diverge sharply from national averages. Banten province illustrates this divergence. Despite its proximity to Jakarta, much of rural Banten comprises low-income households with limited digital literacy, and average rural per-capita monthly expenditure (IDR 1,162,944) is roughly two-thirds of the urban figure (IDR 1,737,427) reported by Badan Pusat Statistik. Consistent with bottom-of-the-pyramid (BoP) research, communication occupies a small and tightly constrained share of the rural household budget, and expenditure is assessed through a strong functional-value lens.

Within this rural setting, Telkomsel ranks third in Branch Serang with a 25.38% market share as of February 2026, behind XL and Indosat Ooredoo Hutchison (IOH). The disparity is anomalous on two counts. First, on a cost-per-gigabyte basis, Telkomsel offers superior data value at comparable price points, ruling out price as an explanation. Second, network quality in the region is objectively equivalent to or better than competitors, ruling out coverage. The anomaly is especially striking for an operator whose national identity is built on superior rural reach. The persistence of a premium perception—the widespread belief that Telkomsel is the most expensive provider even though its prices are now competitive—exemplifies cognitive dissonance, the tendency to discount new information that conflicts with established beliefs (Festinger, 1957), reinforced in rural Banten by tight community networks in which XL has long been framed as the affordable choice.

The problem is both more urgent and more solvable because of two structural shifts in the industry environment. The 2022 transfer of Telkomsel's tower assets to Mitratel opened the same network backbone to competitors, neutralizing coverage as a differentiator (Houngbonon et al., 2025; Wang & Sun, 2022). Concurrently, the Ministry of Communication and Digital Affairs curbed destructive price competition through an informal minimum starter-pack price of IDR 35,000 for 3GB, compressing price differences at the entry level that dominates the rural market. Competitive advantage now rests less on technology or price and more on distribution, community participation, and perception management (Riawan, Pasaribu, & Sutjipto, 2025).

Prior scholarship has examined the constituent elements of this problem—service quality, price fairness, brand image, consumer journey, and rural adoption—but largely in separate streams and rarely in the specific context of an incumbent operator underperforming in a rural emerging market following structural deregulation. This study addresses that gap by integrating four theoretical lenses: the 5A customer journey (Kotler, Kartajaya, & Setiawan, 2017), the BJ Fogg Behavior Model of motivation, ability, and prompt (Fogg, 2009), Cognitive Dissonance Theory (Festinger, 1957), and the bottom-of-the-pyramid perspective on low-income rural consumption. The research is guided by four questions, moving from diagnosis to solution: (RQ1) How has the competitive landscape for Telkomsel changed in rural Banten following the tower consolidation and the 2025 price regulation? (RQ2) What factors influence the provider-selection decisions of low-income rural customers in Banten? (RQ3) How do behavioral segments among rural customers differ, and what are the implications for targeting? (RQ4) What strategic initiatives are most effective for increasing Telkomsel's market share in rural Banten?

2. Method

This study employed a convergent parallel mixed methods design grounded in a pragmatic philosophical stance (Creswell & Creswell, 2023). Qualitative and quantitative data were collected and analyzed in parallel and integrated at the interpretation stage, allowing the internal organizational view of the market share problem to be triangulated against the external customer view (Creswell & Plano Clark, 2018). The design is descriptive-exploratory rather than hypothesis-testing: the objective is to diagnose a specific business problem and to develop an actionable strategy, not to estimate causal parameters.

2.1 Qualitative strand: informants and analysis

Three senior informants with direct involvement in the rural market problem were selected through purposive sampling on the basis of functional knowledge rather than seniority alone (Patton, 2015; Saunders & Townsend, 2016). Each contributed a distinct perspective: a branch-level commercial view (Manager Mobile Consumer), a data-driven analytical view (Officer Mobile Sales Strategy and Analytics), and a field-execution view (General Manager Sales, Business and Partnership). Semi-structured interviews of sixty to ninety minutes covered five themes—competitive dynamics, rural customer behavior, current strategy and execution, organizational constraints, and strategic priorities. Interviews were recorded in Indonesian, transcribed verbatim, and analyzed using deductive thematic analysis following Braun and Clarke (2006), with five overarching themes derived from the literature and research questions, yielding 42 codes and 15 sub-themes. Themes supported by at least two of three informants were treated as supported; member checking confirmed the interpretation without amendment (Birt, Scott, Cavers, Campbell, & Walter, 2016; Lincoln & Guba, 1985).

2.2 Quantitative strand: population, sampling, and measures

The target population was rural prepaid mobile users in Banten province, with low-income users (monthly communication budget IDR 15,000–50,000) as the core segment. A non-probability cluster sampling approach was used because no individual-level sampling frame exists, and respondents were reached through warungs, local markets, community meeting points, and outlet-partner networks across the six clusters (Kabupaten Tangerang, Kabupaten Serang, Kota Cilegon, Pandeglang, Kota Serang, and Lebak). A four-item screening confirmed residence, prepaid SIM use, communication budget, and active smartphone usage. The achieved sample of 347 respondents exceeds the minimum recommended for problem-solving business research and satisfies the expected-cell-frequency requirement for chi-square testing (Malhotra, 2020).

The questionnaire comprised six sections administered in Indonesian: screening, demographics, usage behavior, eighteen five-point Likert items (1 = strongly disagree to 5 = strongly agree), decision factors, and open-ended questions. The Likert items operationalized six constructs drawn from the literature—price perception, perceived value, brand image and trust, social influence and word-of-mouth, the three Fogg components (motivation, ability, prompt), and switching readiness and loyalty—with both positively and negatively worded items to mitigate common method bias (Podsakoff, MacKenzie, & Podsakoff, 2012; Polas, 2025). A pilot of 30 respondents in Kota Cilegon refined item comprehensibility prior to full deployment.

2.3 Analysis and integration

Quantitative data were analyzed through item-level descriptive statistics, between-operator mean comparisons, and cross-tabulation with chi-square tests of independence, reporting Cramér's V as an effect-size measure (Goss-Sampson, 2025). Because sampling was non-probabilistic, p-values are interpreted as indicators of pattern strength within the sample rather than as population inferences. Behavioral segmentation profiled respondents on combinations of age, WiFi availability, top-up behavior, and operator preference; K-means cluster analysis on six ordinally encoded behavioral variables served as a validation check. Microsoft Excel was used for descriptive statistics and cross-tabulation, and JASP for chi-square testing. Integration followed the joint display method (Fetters, Curry, & Creswell, 2013), arranging qualitative and quantitative evidence side by side for each research question to determine convergence, divergence, or expansion (Creswell & Inoue, 2024).

3. Results

Results are presented in four stages: the sample profile, item-level perception and behavior patterns, cross-tabulation findings, and behavioral segmentation, followed by the joint display integrating the qualitative and quantitative strands.

3.1 Sample profile

The 347 respondents were distributed across all six clusters (Table 1). The demographic profile confirms the rural low-income focus: 75.8% were aged 18–34, 70.9% had completed senior secondary education (SMA/SMK), and the sample was 55% female. Home WiFi penetration reached 69.4% (60.2% private WiFi plus 9.2% informal RT/RW Net), notably higher than is commonly assumed for rural Indonesia and consistent with informant accounts of informal community internet reshaping the market. Self-reported Telkomsel usage in the sample (58.8%) substantially exceeds the actual 25.38% branch market share, reflecting self-selection bias; consequently, aggregate shares are treated as sample descriptions, while between-operator comparisons—which rely on within-group differences—remain analytically valid.

Table 1: Sample distribution and key demographics (n = 347)

3.2 Perception and behavior patterns

Item-level statistics (Table 2) show the highest mean on the brand-exclusivity item (D9, "Telkomsel is for more affluent individuals", M = 3.80), quantitatively confirming the qualitative perception that Telkomsel is a premium brand. Loyalty items (D17, D18) ranked among the highest, while switching intent (D16, M = 2.71) was the lowest, indicating limited active switching. The strategically decisive pattern emerges in between-operator comparison: on the premium-perception item D1, non-Telkomsel users reported higher means (IOH ≈ 3.79, XL ≈ 3.53) than Telkomsel users (3.24). Those with no direct experience of Telkomsel's current pricing hold the strongest belief that it is expensive—precisely the belief that deters trial—providing empirical support for the cognitive-dissonance mechanism. Direct experience, conversely, raised trust and loyalty: Telkomsel users scored higher on brand-trust items (≈ 3.80) than XL (3.34) and IOH (3.36) users, and on loyalty items (≈ 3.70) than XL (3.26) and IOH (3.46) users. The premium perception did not vary significantly across monthly spending brackets, indicating it is widely held across the rural sample rather than confined to the lowest-spending group.

Table 2: Selected item-level descriptive statistics (n = 347)

3.3 Cross-tabulation analysis

Four associations were statistically significant (Table 3): operator choice by region, by age group, by purchase location, and—most strongly—by prior Telkomsel use (V = 0.662). The generational pattern is strategically central: Telkomsel's share falls to 46.7% in the 12–17 cohort against 71.7% at 35–44, a 25-point gap (χ² = 36.93, p = 0.001) that confirms the youth segment is underserved. The WiFi-disruption pattern was equally clear: 66.6% of respondents reported buying less mobile data after adopting home WiFi (47.6% much less, 19.0% slightly less), with only 1.4% buying more—evidence of a structural, not perceptual, contraction in rural mobile-data demand. By contrast, 87.9% of respondents described operator choice as made independently, against an informant estimate of 60% family influence, suggesting family influence operates as implicit normative pressure rather than explicit decision authority (social-influence item mean 3.33).

Table 3: Chi-square tests of independence — summary

Note. *** p < 0.001; ** p < 0.01; n.s. = not significant. χ² = chi-square statistic.

3.4 Behavioral segmentation

Four behavioral segments emerged from combinations of age, WiFi availability, top-up behavior, and operator preference (Table 4). K-means clustering on six ordinal variables produced a directionally aligned four-cluster solution (silhouette ≈ 0.213; cluster shares of ~28/23/22/26% against observational estimates of ~33/25/22/20%), supporting the segmentation. Young Swing Voters (S3, ~22%) combine the highest acquisition potential with the largest perception-behavior gap and are identified as the priority acquisition target, given the risk of multi-year competitor lock-in.

Table 4: Four behavioral segments in the rural Banten sample

3.5 Integration of qualitative and quantitative findings

The joint display (Table 5) shows strong convergence across the major themes. Three findings converged most strongly: the premium-perception gap between users and non-users, the 66.6% WiFi-driven reduction in mobile-data buying, and the 25-point generational gap in Telkomsel share. The apparent divergence on family influence (60% claimed versus 11.6% explicit) resolved, on integration, into a single mechanism—implicit normative pressure—that neither strand could have identified alone.

Table 5: Joint display: selected qualitative–quantitative integration

4. Discussion

The integrated evidence answers the four research questions and points to a single overarching conclusion: the Telkomsel market share gap in rural Banten is not, in absolute terms, a network, price, or product-quality problem, but a perception, behavior, and execution problem. Five interlocking drivers account for the gap.

First, the basis of competition has shifted decisively away from infrastructure and price (RQ1). Tower sharing neutralized Telkomsel's long-standing coverage advantage, and informal price regulation compressed entry-level price differences, consistent with evidence that infrastructure-based advantages are temporary once sharing arrangements take hold (Houngbonon et al., 2025; Wang & Sun, 2022) and that deregulation shifts competition toward service and experience (Hildebrand & Wiewiorra, 2023). Crucially, customers have not yet internalized this change: premium perception remains high even as the pricing reality has equalized.

Second, provider selection is driven by five socially mediated forces rather than any single factor (RQ2): acknowledged price sensitivity, implicit family and community normative pressure, the user/non-user perception gap, the structural impact of home WiFi, and generational divergence. This multi-factor structure echoes Maity and Singh's (2020) finding that market development in low-income Asian segments depends on the joint operation of awareness, accessibility, affordability, and acceptability, and implies that a single campaign cannot move the market.

Third, behavioral segmentation reveals four distinct segments requiring differentiated treatment (RQ3), of which Young Swing Voters are the most underserved and strategically important. Fourth, the most effective interventions are grassroots community programs (Kampung Telkomsel and Kampung byU) reinforced by competitive outlet incentives, whereas mass-media spending converts poorly in this market (RQ4)—a pattern aligned with community-engagement and local-knowledge perspectives on hyperlocal marketing (Ganapathy, 2024; Sharma, Bhardwaj, Farheen, & Agarwal, 2025). Fifth, execution of these levers is constrained by branch-level authority, so corporate enablement in budget and delegated decision rights is a structural prerequisite.

Theoretically, the study extends Cognitive Dissonance Theory to rural emerging-market telecommunications, showing that non-experiential beliefs are highly resistant to factual updating and are reinforced where the belief is shared across a community rather than held individually (Schmidtke, Rundle-Thiele, Kubacki, & Burns, 2020). It extends the BJ Fogg Behavior Model to provider switching in low-literacy rural contexts, where motivation, ability, and prompt are each constrained (Jyothy, Kolil, Raman, & Achuthan, 2024), and enriches the bottom-of-the-pyramid literature by demonstrating that distinct behavioral segments emerge even within a relatively homogeneous rural population (Fahmi & Arifianto, 2022). Managerially, the findings imply that perception change in such markets is best achieved through direct experience rather than messaging, that defensive product design is required as home WiFi erodes mobile-data ARPU, and that softer levers—community presence, channel execution, and brand image—now determine competitive advantage in deregulated rural telecommunications markets.

Several limitations qualify these conclusions. The survey over-represents Telkomsel users (58.8% versus 25.38% actual share), so aggregate statistics describe the sample rather than the population, although between-group comparisons remain robust. The single-branch scope limits geographic generalisability; the cross-sectional design precludes causal inference; self-reported attitudes carry social-desirability risk; and the qualitative strand reflects only internal stakeholders. Future research should pursue longitudinal validation, multi-stakeholder extension to competitors and community leaders, cross-regional comparison, and field experiments contrasting experiential and communication interventions to test the dissonance mechanism directly.

5. Conclusion

This study investigated why Telkomsel, the national mobile market leader, ranks third in rural Banten despite competitive pricing and coverage. Using a convergent parallel mixed methods design, it established that the 23-percentage-point gap between Telkomsel's national share (48.6%) and its Branch Serang share (25.38%) is primarily a perception, behavior, and execution problem rather than a network or price problem. Five interlocking drivers explain the gap: a persistent premium perception strongest among non-users, behavioral inertia reinforced by family and community norms, structural ARPU decline from home WiFi adoption, an underserved youth segment, and constrained branch-level execution authority.

All five drivers are addressable through a phased implementation of evidence-based initiatives. The data identify and validate the most effective levers—grassroots community programs such as Kampung Telkomsel, competitive outlet incentives, community-leader involvement, and a youth-specific product offering—integrated through TOWS, STP, and the 7P marketing mix into a branch- and corporate-level roadmap. The youth segment is the priority acquisition target, because a customer won or lost at this age tends to remain with that operator for years. Beyond Telkomsel, the study offers a transferable diagnosis for incumbent operators facing similar erosion in deregulated rural emerging markets: once infrastructure and price are equalized, durable advantage is built through perception management, community engagement, and disciplined channel execution.

Author Contributions: Conceptualization, R. A. S. Akbar; methodology, R. A. S. Akbar; investigation, R. A. S. Akbar; formal analysis, R. A. S. Akbar; writing—original draft preparation, R. A. S. Akbar; writing—review and editing, I. A. Zaim and J. S. Mussry; supervision, I. A. Zaim and J. S. Mussry. All authors have read and agreed to the published version of the manuscript.

Funding: This research received no external funding.

Conflicts of Interest: The corresponding author was affiliated with the subject organization during the research, which provided access to internal data used in aggregate form. Procedural safeguards—standardized survey instruments, multi-perspective interviews, and external academic supervision—were applied to mitigate interpretive bias. The authors otherwise declare no conflict of interest.

Informed Consent Statement / Ethics Approval: The research adhered to the ethics requirements of the MBA program at Institut Teknologi Bandung. All survey respondents gave informed consent prior to participation, and all responses were anonymized and used in aggregate. The three interview informants provided written consent for recording, transcription, and attribution of their responses. Use of confidential corporate data was authorized in writing by company management.

Data Availability Statement: The aggregate survey data and anonymized interview summaries supporting the findings of this study are available on request from the corresponding author. Internal corporate data are subject to confidentiality restrictions.

Declaration of Generative AI and AI-assisted Technologies: This study has not used any generative AI tools or technologies in the preparation of this manuscript.