Abstract
Self-order kiosks (SOKs) are increasingly common in restaurants, yet many older adults experience them as frustrating rather than convenient because designs are not aligned with age-related needs. Prior research has largely centred on usability problems, while older users’ emotional and cognitive challenges have received less attention. To provide a fuller account of challenges and coping responses, this study combined empathy mapping with thematic analysis to capture what participants said, thought, felt, and did during interaction. We conducted a qualitative case study in Malaysian restaurants, integrating kiosk observation with focus-group discussions. Thirty older adults participated across eight focus groups. Full-sample familiarisation was completed for all cases, followed by a mechanism-focused analysis of a purposive subsample of 15 cases. Three themes emerged: (a) anxiety and queue pressure (Hear × Feel) shaping avoidance and help-seeking; (b) affordance clarity and legibility (See → Do) reducing unnecessary load and error loops; and (c) payment complexity moderated by facilitating conditions (intrinsic load × support) differentiating bail-outs from recovery. Multi-step payments increased intrinsic load and raised bail-outs to the counter, while text and visual clutter elevated extraneous load, contributing to mis-taps and restart loops. Recovery was more likely when help affordances were visible or staff were nearby.
We provide a theory-driven conceptual mapping linking cognitive load mechanisms and STAM/UTAUT determinants to empathy-map quadrants, where quadrant interaction patterns (for example, Hear × Feel under queue pressure) explain avoidance and help-seeking. Findings are bounded to Malay/English SOK flows and Malaysian restaurant contexts, with strongest transferability to comparable self-service technologies.
Plain Language Summary
Self-order kiosks (SOKs) are becoming more common in restaurants, but many older adults find them difficult and frustrating. These challenges go beyond just learning how to use the screens—they also involve emotional stress like confusion, anxiety, and feeling overwhelmed.
To better understand these experiences, researchers interacted with 30 older adults in Malaysia through group discussions. Instead of asking questions, they used empathy mapping, which looks at what people say, think, feel, and do. This gave a fuller picture of how older adults interact with kiosks in real situations.
The study found three main issues:
Emotional Frustration—Many felt anxious, confused, or lacked confidence when using kiosks.
Design Problems—Issues included small text, hard-to-use touchscreens, and unclear steps for ordering and paying.
Suggested Improvements—Participants recommended changes like larger fonts, easier instructions, and voice guidance.
This research highlights ways to improve kiosk design and demonstrates how empathy mapping can help researchers understand users’ experiences. The findings can assist businesses, designers, and policymakers in developing more user-friendly and inclusive technology for older adults.
Introduction
The integration of self-order kiosks (SOKs) in restaurant settings has been widely recognised as a means to enhance operational efficiency and reduce reliance on human labour (Baba et al., 2023). However, while these technological advancements benefit younger, tech-savvy consumers, they often pose significant usability and accessibility challenges for older adults. Research indicates that older adults frequently struggle with complex interface designs, small text sizes, and lack of immediate support, leading to frustration and avoidance of self-service technology (SST; Hashim et al., 2023; Vozza, 2024). These challenges are not merely technical but also encompass emotional and behavioural responses influenced by cognitive load, situational stress, and prior digital literacy (Nam et al., 2023).
Although prior work has documented older adults’ barriers with SSTs (e.g., Leite et al., 2022), we foreground recent (last 5 years), peer-reviewed HCI/gerontechnology literature and prune peripheral or non-scholarly sources. We position our study to interrogate not only what barriers exist, but how emotional and cognitive mechanisms shape avoidance and help-seeking at SOKs.
We address this gap through three research questions: (a) Which usability and accessibility challenges do older adults face at SOKs in restaurant settings? (b) How do personal, technical, and situational factors shape their emotions and behaviours during these interactions? (c) What insights from empathy mapping can inform more inclusive SOK design? We conducted a qualitative multiple-session case study in Malaysia using focus-group discussions (FGDs) with 30 older adults and in-situ observation. We seek analytical generalisation rather than statistical inference. Evidence highlights how touchscreen sensitivity, lack of auditory feedback, and intricate menu navigation heighten cognitive burden, reduce confidence, and increase dependence on staff support (Liu et al., 2022; Vermeulen et al., 2013; Zulkafly et al., 2021).
Furthermore, we integrate empathy mapping with a theory-informed lens. We link the four quadrants to determinants discussed in ageing-focused acceptance models and technology-use frameworks, and interpret barriers using cognitive load concepts. This helps explain avoidance and help-seeking behaviours in context. The approach is valuable for accessibility research because it captures non-verbal cues and real-time frustrations often missed by standard thematic analysis (Leite et al., 2022). Findings underscore the need for larger base text, streamlined navigation, clear payment feedback, bilingual labels, and visible help affordances—items directly supported by participant evidence. Personalised assistance technologies discussed in the literature (Shin et al., 2023) are acknowledged as future work, separate from the evidence-based implications presented here. This paper contributes: (a) a conceptual mapping between empathy-map quadrants and adoption/usage determinants relevant to older adults; (b) evidence of quadrant interaction patterns, for example, Hear × Feel under queue pressure, that predict bail-out vs. recovery; (c) an evidence-to-implication traceability linking quotes/observations to specific design recommendations. Findings are bounded to Malaysian quick-service contexts and common SOK brands/flows in Malay/English; transferability is strongest to similar SOK environments.
Literature Review
Empathy mapping has emerged as a useful qualitative device in HCI/gerontechnology for understanding users’ perceptions, emotions, and actions during interaction (Gray, 2017; Leite et al., 2022; Samengon et al., 2024). In this review, we foreground how empathy maps can be read through a theory-informed lens to interpret older adults’ SOK experiences.
As illustrated in Figure 1, structuring data into the four quadrants (what users say/think/feel/see/hear/do) surfaces patterns that can be linked to known determinants of technology use (e.g., perceived difficulty, social pressure, and support conditions) and to cognitive burden created by visual clutter or multi-step flows.
Empathy maps structure.
Prior empirical work indicates that empathy-led service and interface refinements, such as clearer affordances, improved legibility, and better recovery prompts, can reduce user errors and strengthen confidence in self-service settings (Henderson et al., 2024; Leite et al., 2022). Beyond interface changes, the empathy-mapping process helps surface emotion-laden frictions that trigger bail-out or help-seeking in public self-service encounters, especially for older and novice users.
In sum, empathy mapping is valuable not only for documenting barriers but also for providing a structured interpretive lens that connects observed behaviours with underlying emotional and cognitive mechanisms, particularly relevant to older adults at SOKs (Gray, 2017; Leite et al., 2022; Samengon et al., 2024).
Usability and Accessibility Enhancements for Older Adults
Older adults and people with disabilities encounter overlapping access and usability barriers at SOKs; small text, low contrast, complex flows and limited multimodal feedback frequently discourage engagement (Czaja, 2019). While accessibility and usability share principles, accessibility attends to perceptual and motor constraints, whereas usability focuses on task efficiency, effectiveness and satisfaction in the user interface.
Physical and visual design factors materially shape SOK use among older adults, including reach, screen readability, lighting, and the salience of on-screen affordances, particularly under time pressure (Nam et al., 2023). In quick-service contexts, service design cues and inclusive wayfinding also shape whether users persist independently or shift toward staff assistance (Henderson et al., 2024). Together, this points to a combined interface-and-context problem: design clarity matters, but it becomes critical when public pressure compresses decision time.
Interface readability (font size, contrast), ergonomic touch targets, and simple, linear flows can mitigate error-prone steps and reduce abandonment (Wang et al., 2024). Studies likewise note that payment steps and health-related kiosk tasks are hindered by perceived risk and complexity among older adults (Kim et al., 2022), with low digital literacy exacerbating breakdowns (Martins et al., 2025).
Across recent work, a consistent theme is that older adults’ kiosk difficulties are not only about “not knowing how,” but about interaction costs accumulating across steps. The literature agrees on the existence of recurring pain points, yet it is less precise about how these costs escalate into avoidance versus persistence. This motivates our focus on mechanism-level explanations rather than additional enumeration of interface defects.
Emotional and Behavioural Dynamics in Older Adults Interaction With SOKs
The emotional and behavioural dynamics of SOK use in later life include anxiety under time pressure, uncertainty after errors, and reliance on proximate assistance. These patterns shape avoidance and help-seeking responses and are salient in service contexts that privilege speed. Practice-oriented discussions similarly stress the need to account for users’ feelings and expectations to sustain engagement (Henderson et al., 2024; Leite et al., 2022).
We use an empathy map as a structured interpretive device to organise context cues, internal states, and observable coping behaviour during kiosk interaction, including what users see and hear in the environment, what they think and feel, and what they say and do. In recent applied work, empathy mapping has been used to surface context-dependent frictions and guide concrete service and interface adjustments, especially when social context and support conditions shape experience (Henderson et al., 2024; Leite et al., 2022). In the present study, this structure anchors our interpretation of quotes and observations by supporting a cue → emotion/cognition → coping action → outcome reading.
Taken together, recent findings suggest that emotional responses, especially anxiety under time and social pressure, are central to kiosk non-use and early abandonment. However, most studies report these emotions as outcomes rather than tracing their triggers and behavioural consequences step by step. This is where empathy mapping is useful, because it allows us to connect specific situational cues to internal states and then to coping actions.
Role of Human Support and Integrated Assistance Systems for Older Adults
Proximate human support (concierge staff, visible assistance points) remains pivotal during onboarding and recovery from errors, making transitions to self-service more intuitive for older adults (Hashim et al., 2022).
In tandem, service-flow refinements (clearer entry points, simplified branching, consistent payment steps) encourage repeat SOK use (Baba et al., 2023).
Built-in assistance, on-screen help, kiosk-side signage, audio prompts reduce frustration and sustains user confidence, particularly at payment and confirmation stages (Leite et al., 2022).
Taken together, a blended model of attentive staff plus clear, recoverable kiosk flows supports completion without undermining independence for older adults.
The literature largely supports the idea that perceived ability and confidence shape whether older adults persist with self-service technologies. Still, there is limited detail on what design cues are interpreted as “safe to proceed” versus “risky to attempt”, especially during payment and confirmation stages. We therefore treat self-efficacy as context-sensitive and observable through empathy-map patterns that reveal when confidence drops and what recovery support is sought.
Technological Innovation for Cognitive and Physical Accessibility
Recent work emphasises cognitive and physical accessibility in self-service contexts, noting that older adults often struggle more with new multi-step digital tasks than with familiar ones (Vozza, 2024).
Consistent concerns arise in domain-specific kiosks, where perceived risk, terminology, and task complexity depress successful use among older adults (Kim, 2023).
Related studies highlight how interface simplification, progressive disclosure, and multimodal feedback can make kiosk interactions more intuitive and efficient for a wide user base (Lee & Lee, 2024).
Building on these findings, we read technology-use patterns alongside acceptance constructs and cognitive demands, given their direct influence on uptake and sustained usage.
Recent studies underline the importance of facilitating conditions, including staff availability, on-site guidance, and recovery affordances. Yet existing work often treats support as a general moderator rather than identifying what specific forms of support are perceived as legitimate, timely, and non-embarrassing. Our empathy-map approach captures these practical distinctions and allows the implication section to specify which support types work for which breakdown moments.
Besides, recent SOK/SST studies involving older users repeatedly report surface-level friction points such as legibility, menu complexity, ambiguous confirmations, and reliance on staff at critical steps (for example, payment). That part is not disputed. What is less consistently articulated is the mechanism that turns a friction point into either (a) a recoverable stumble or (b) immediate abandonment and a return to the counter. In other words, the same design flaw can lead to different behavioural outcomes depending on situational pressure (queues, noise, being watched), perceived risk (especially at payment), and the visibility of recovery cues (help affordances, staff proximity). To move beyond a descriptive “issue list,” we therefore organise the next section around constructs that are repeatedly used to explain older adults’ technology use in service settings: anxiety/technostress under time pressure, self-efficacy, and facilitating conditions, interpreted through cognitive load distinctions (intrinsic vs. extraneous). These constructs also map cleanly onto what empathy mapping can capture in a structured way (what users see and hear in context, what they feel and think, and what they do next) (Shin et al., 2023; Wu et al., 2023).
To reflect the current state of research, this review prioritises peer-reviewed studies from roughly the last 5 years that examine (a) older adults’ interaction barriers with self-service kiosks and closely related self-service technologies, (b) the role of public and time-pressured contexts in shaping affective responses such as anxiety, embarrassment, and low confidence, (c) facilitating conditions such as on-site support and help affordances that determine whether users persist or withdraw, and (d) payment and confirmation steps as risk-intensive moments that amplify uncertainty and error costs. Earlier work is retained only when it is foundational for the conceptual framing (for example, core accessibility principles or established technology-use constructs), while older studies that repeat well-established usability issues without extending the explanation are reduced. This positioning clarifies the gap addressed by the present study: not simply restating known kiosk usability problems, but explaining how those problems escalate into avoidance, help-seeking, or bail-out in real restaurant settings, and under what conditions recovery becomes possible.
Theory Alignment and Analytic Expectations (Why These Propositions, and What Each Variable Means)
In this study, propositions are not introduced as “hypotheses to be tested,” but as theory-anchored expectations that guide how empathy-map cues are interpreted and grouped into higher-order patterns. The literature on older adults’ technology use consistently points to four determinants that matter in self-service settings: (a) situational anxiety under time and social pressure, (b) self-efficacy, (c) facilitating conditions (availability and visibility of assistance), and (d) cognitive load created by interface design and multi-step flows. In kiosk interaction, these determinants rarely appear in isolation. A queue behind the user is not merely a “context variable”; it changes what the user hears (ambient noise, comments from others), which can elevate worry and perceived error cost, and in turn shifts coping behaviour toward rushing, stepping aside, or seeking help.
We therefore define our focal constructs in the specific SOK context as follows. Queue-pressure anxiety refers to the emotional and cognitive strain triggered by perceived time pressure and being observed by others during ordering. Self-efficacy refers to the user’s perceived capability to complete ordering and payment without making an error that has social or financial consequences. Facilitating conditions refer to the practical availability of support that allows recovery from stalls (for example, visible staff assistance or clear on-screen help affordances). Cognitive load is treated as an interaction cost that accumulates when the interface demands effortful perception (legibility), decision-making (menu branching), and confirmation (payment and order finalisation). These constructs are used to interpret the empathy-map quadrants not as descriptive bins, but as a way to connect external cues (See, Hear) to internal states (Think, Feel) and then to coping actions (Do) and outcomes (recovery vs. bail-out).
This framing also clarifies the theoretical contribution: rather than re-listing known kiosk usability issues, we explain how they escalate into abandonment or, alternatively, how recovery becomes possible when facilitating conditions and confidence cues are present. This is consistent with the manuscript’s cue → emotion/cognition → coping action → outcome reporting structure used in the Findings.
PR1 (Queue pressure and social exposure: Hear × Feel → anxiety → avoidance/help-seeking).
Variables: situational pressure (queue noise, feeling watched, time urgency), anxiety (worry about slowing others, fear of mistakes), outcome (avoidance, outsourcing to staff, or premature bail-out). Prior kiosk and SST research in service environments has repeatedly highlighted that anxiety is not only an internal trait, it is triggered by context, particularly when systems are framed as “fast” and public. In older adults, this anxiety is tightly coupled with risk perceptions and embarrassment costs (especially when errors are visible). In our data, this mechanism will be observable as Hear cues (queue noise, comments from others, perceived impatience), Feel cues (stress, shame, panic), followed by Do actions (step back, ask staff, switch to counter) (Kim & Park, 2022; Nam et al., 2023).
Rationale: Queue noise and being observed heighten perceived error cost, which lowers confidence and increases avoidance or help-seeking, especially when system feedback is weak. Recent SST/kiosk work links public time pressure to anxiety-driven avoidance and help-seeking among older adults, suggesting this pathway is not idiosyncratic to our setting.
PR2 (Presentation quality: See → Do → extraneous load → error loops).
Variables: legibility/affordance clarity (font size, contrast, icon meaning, touch targets), extraneous load, behavioural outcome (mis-taps, restarts, time loss). Older adults’ kiosk breakdowns often follow a predictable loop: unclear visual hierarchy or ambiguous icons increase search time, then small touch targets increase error likelihood, and errors increase stress, which further reduces accuracy. In our data, this will be observable as See cues (small text, crowded menu, unclear back/confirm), paired with Do cues (repeated tapping, restarting, scanning the screen, pointing). The analytic claim is not that “small text is bad” but that it creates a measurable behavioural pattern: error loops and delayed recovery under pressure (Henderson et al., 2024; Leite et al., 2022).
Rationale: When text and affordances are hard to parse, users increase exploratory tapping and backtracking, which raises extraneous cognitive load and creates error loops.
PR3 (Confidence to proceed: self-efficacy as a buffer linking Think/Feel to persistence).
Variables: self-efficacy, emotional response (reduced anxiety), behavioural outcome (persistence, completion). Self-efficacy is consistently treated as a key determinant in older adults’ technology acceptance: users persist when they feel able to correct mistakes and predict what the system will do next. In empathy maps, self-efficacy is most visible not as a single statement, but as a pattern: users verbalise certainty (“I know what to press”), move more decisively, and recover from minor slips without abandoning the task. In our data this will be observable as Think cues (certainty vs. confusion) and Do cues (steady progression vs. hesitation) (Shin et al., 2023).
Rationale: Prior exposure and perceived competence shift behaviour from “stall and withdraw” to “explore and persist”, making self-efficacy observable in the coping actions recorded in the empathy map.
PR4 (Recovery support: facilitating conditions convert stalls into recovery rather than bail-out).
Variables: facilitating conditions (staff proximity, help button, signage, clear “undo/back,” clear feedback), outcome (recovery vs. bail-out). Extended UTAUT work applied to older adults using terminals shows that facilitating conditions are not abstract; they are experienced as “there is a clear next step and someone or something will help if I get stuck”. In empathy maps, this will appear when users explicitly reference the availability of help (or its absence) and when “stall moments” end either with recovery or with abandonment (Wu et al., 2023).
Rationale: Visible, low-friction assistance enables recovery without undermining autonomy, so facilitating conditions function as a moderator of stall-to-completion outcomes. Extended UTAUT findings consistently show facilitating conditions predict persistence and recovery behaviour during breakdown moments, which aligns with our quadrant interaction reading.
PR5 (Cumulative fragility: extraneous load + intrinsic load → resource depletion → bail-out at payment).
Variables: extraneous load (clutter/legibility), intrinsic load (multi-step payment and confirmations), outcome (bail-out, handover to staff, refusal to use kiosk next time). Payment is a high-risk step for older users because the cost of “getting it wrong” feels larger, and the system feedback is often less transparent than earlier browsing steps. When poor presentation and complex branching co-occur at payment, users’ coping capacity is exceeded and abandonment becomes rational. In empathy maps, this will show as an escalation sequence: See (multiple payment options), Think (uncertainty about selection/confirmation), Feel (fear), then Do (stop, seek staff, move to counter) (Kim & Park, 2022; Wu et al., 2023).
Rationale: Co-occurrence of legibility burden and multi-step payment uncertainty increases the total interaction cost, making bail-out more likely than when only one friction is present.
These propositions are not hypotheses for statistical testing. They are theory-anchored analytic expectations that guide how empathy-map cues are interpreted across quadrants and translated into behavioural outcomes. Recent studies on older adults and self-service technologies consistently identify anxiety under social/time pressure, self-efficacy, facilitating conditions, and interaction burden as key determinants of persistence or withdrawal (Nam et al., 2023; Shin et al., 2023; Wu et al., 2023).
In SOK settings, these determinants do not operate independently. The same interface issue can lead to different outcomes depending on context. Queue pressure, perceived social scrutiny, and weak recovery cues can amplify uncertainty, increase perceived error cost, and push users towards help-seeking or abandonment, especially at high-risk stages such as payment confirmation (Kim & Park, 2022; Zhu & Husin, 2022; Zygouris et al., 2022). This is why the empathy-map structure is used here to connect environmental cues (See/Hear), internal processing (Think/Feel), and enacted response (Do) in a single cue-to-outcome pathway.
Accordingly, the contribution is explanatory rather than descriptive. Instead of re-listing familiar usability problems, this study specifies how friction escalates into avoidance, bail-out, or recovery, and under which support conditions recovery becomes likely. This mechanism-oriented framing also aligns the Findings and Implications sections with theory-driven interpretation (Shin et al., 2023; Silva et al., 2021; Wu et al., 2023).
Methodology
Research Design
This study employed a qualitative, multiple-session case study to examine older adults’ use of self-order kiosks (SOKs) in Malaysian quick-service restaurants. Data were generated through focus-group discussions (FGDs) with 30 older adults and in-situ kiosk observation, providing both articulated experiences and situated behaviours relevant to usability and accessibility. We seek analytical generalisation from rich cases rather than statistical inference; full procedures and safeguards are detailed below. We conducted eight focus-group sessions (S1–S8), each treated as a discrete session for analysis and saturation tracking.
Sampling Procedure
A purposive sampling strategy recruited 30 older adults across eight FGDs to balance breadth with depth and to approach theme-level saturation. Inclusion criteria: aged ≥60 years; prior visit to a quick-service restaurant in the last 6 months; ability to communicate in Malay or English; self-rated ability to read on-screen text with/without corrective aids. Exclusion criteria: diagnosed severe cognitive impairment; acute visual/motor conditions preventing kiosk use. This sample as shown in Table 1 comprised 30 older adults recruited in Selangor and Negeri Sembilan states in Malaysia. Ages were distributed across 60 to 64 (27%), 65 to 69 (30%), 70 to 74 (27%) and 75+ (17%). Women made up 57% and men 43% of participants. Education levels included primary/lower secondary (30%), upper secondary (37%) and tertiary (33%). Most were retired (60%), with 27% employed and 13% homemakers/other unpaid. Prior SOK experience varied: 40% had never used a kiosk, 37% had used one 1 to 3 times, and 23% had used one 4+ times. Self-reported visual limitations were present in 37% and motor limitations in 20%. The primary interface language used was Malay (50%) or English (43%), with a small number using both during the session (7%). Quoted participant IDs reflect the 15-case analytic subsample; full recruitment and participant characteristics are reported for all 30 participants.
Participant Characteristics (n = 30).
We approached 68 older adults across churches, online community groups, a local university, family networks and public places in Selangor and Negeri Sembilan; 44 were screened, 32 met inclusion criteria, and 30 consented and participated in eight focus-group discussions. Groups were scheduled across varied sites and times to capture plausible variation in ambient conditions while maintaining feasibility and participant comfort. Each FGD (60–90 min) followed the moderator guide aligned to empathy-map quadrants and concluded with a brief in-situ kiosk walkthrough where feasible.
The semi-structured focus group discussions were strategically guided by the empathy map framework, which provided a structured yet flexible lens to explore participants’ sensory, cognitive, and emotional engagements with self-order kiosks (SOKs) in fast-food settings. The sessions began with two warm-up questions designed to ease participants into the discussion by reflecting on their general relationship with technology and their prior experiences using SOKs. This helped establish rapport and contextual grounding before delving into the core dimensions of the empathy map: what participants hear, see, think and feel, say and do, as well as their perceived pains and gains.
This approach enabled the research team to systematically capture a holistic view of user experiences, including environmental cues, internal thought processes, verbal expressions, and behavioural responses. By eliciting insights across these domains, the method facilitated rich, nuanced discussions that surfaced both implicit and explicit reactions to SOK use.
To enhance methodological rigour and reduce reliance on retrospective self-reporting, the FGDs also incorporated real-time usability observation. Prior to the main discussion, participants were briefed on the research objectives and the structure of the empathy map, after which they were invited to interact with actual SOK interfaces in a controlled setting. This embedded activity allowed researchers to observe spontaneous user behaviours, emotional cues, and problem-solving strategies in situ. The triangulation of observed interactions with participants’ verbal narratives strengthened the reliability and depth of the findings, as recommended by Farsani (2022), by revealing gaps between what users say and what they actually do.
Ethical Considerations
This study adhered to ethical guidelines for research with human participants, including informed consent, confidentiality and secure data handling. Consent explicitly covered verbatim quotation of de-identified excerpts, and all participants were anonymised as P01 to P30; reported quotations are drawn from the purposive 15-case analytic subsample. Accessibility accommodations (font size, seating, pace) were provided where needed, and participants could refrain from any activity without penalty (Vermeulen et al., 2013).
Data Analysis
Preliminary Observations Through Empathy Map Dimensions
We conducted a multi-stage analysis that prioritised inductive coding before any quadrant assignment. FGDs were audio-recorded, transcribed verbatim and organised in NVivo. Two researchers performed open coding to surface patterns; only then were codes related to empathy-map quadrants to structure cross-case comparison. In this sequence, the empathy map functioned as a sensitising device, not a forcing template. A summary of quadrant-linked observations with participants’ responses is presented in Table 2.
Summary of Empathy-Map Observations With Participants’ Responses.
The collected data underwent a multi-stage thematic analysis, following Braun and Clarke’s (2006) framework: (a) Familiarisation; (b) systematic coding in NVivo; (c) development of higher-order themes around usability challenges, emotional responses and design recommendations; and (d) refinement through analyst triangulation and consensus. Two analysts coded independently and then held consensus meetings to compare interpretations and agree theme boundaries. Member checking with selected participants confirmed the fairness of our summaries. These steps produced the overarching themes reported below, offering a richer understanding of participants’ experiences with SOKs (González-Bañales & Ortiz, 2017; Liu et al., 2022).
Coding and Categorisation
User feedback analysis on SOKs revealed recurring usability challenges, emotional responses, and behavioural adaptations among older adults. The qualitative data were analysed using thematic analysis, following Braun and Clarke’s (2006) six-phase framework, which provided a systematic approach for identifying, analysing, and reporting patterns in qualitative data.
Data were transcribed and imported into NVivo software for systematic coding, ensuring a transparent and replicable process. An initial open coding phase was conducted independently by two researchers to generate preliminary codes (Table 3), which were then refined into axial codes (Table 4) to establish relationships between usability challenges and user experiences. In the initial open coding as shown in Table 3, all names have been anonymised to maintain confidentiality. Technical issues, including touchscreen sensitivity (Codes 1, 3, 16) and small font sizes (Codes 2, 25, 36), are commonly reported, highlighting the necessity for enhanced interface design. Users frequently report dissatisfaction stemming from these constraints, which hinder navigation (Codes 6, 9, 23, 26), order confirmation (Codes 5, 38), and payment method selection (Codes 8, 28, 39). Research indicates that improving the user interface can markedly enhance customer satisfaction and usability, which is essential for the adoption of SOKs across different contexts (Hamid et al., 2021; Ishak et al., 2021; Xavier et al., 2023).
Preliminary Codes.
Axial Codes and Main Themes.
Note. Axial categories are not mutually exclusive and may contribute across themes; row placement reflects dominant mechanism alignment within this analysis.
Accessibility presents a significant issue, especially for older adults and visually impaired individuals who encounter difficulties with small text (Codes 2, 20, 25, 36). Research highlights the necessity of designing kiosks that cater to varied user requirements, such as auditory feedback and enlarged text options, to improve usability across all demographic groups (Dela Peña & Aquino, 2023; Sandnes et al., 2011). The presence of cognitive overload stemming from intricate interfaces and extended queues (Codes 12, 17, 18, 26, 32) highlights the imperative for more straightforward instructions and enhanced user support (Codes 10, 19, 22, 27). This is consistent with research indicating a preference for human interaction when technology does not fulfil user expectations (Ishak et al., 2021).
Users have indicated a preference for expedited transactions and an improved learning experience (Codes 22, 30, 31, 32, 37). The incorporation of various payment methods, such as e-wallets (Code 39), is crucial for enhancing user confidence and satisfaction with SOKs. Research demonstrates that the perceived usefulness of kiosks, especially in minimising waiting times, significantly influences user acceptance (Hamid et al., 2021; Kim & Qu, 2014). Addressing usability and accessibility issues can enhance the overall customer experience and promote broader adoption of SOKs in the fast-food industry and beyond (Ishak et al., 2021).
The analysis of user feedback identifies key areas for improvement in SOK design, such as technical enhancements, accessibility features, and streamlined user interfaces. These adjustments are essential for enhancing user satisfaction and ensuring that SOKs address the varied needs of their user base. To enhance analytic rigour, two analysts used independent coding followed by consensus, documented decisions in analytic memos, and tracked saturation by theme; no new codes emerged for the first two themes after the session 7 FGD, with only a minor elaboration in Theme 3 during session 8.
Findings and Discussion
We organise the findings using the cue → determinant → cognitive load → outcome chain outlined in Theory Alignment and Analytic Expectations (Why These Propositions, and What Each Variable Means section. Initial coding yielded 39 first-order codes, consolidated into 12 axial categories and three aggregate themes (Table 4). Across these themes, we revisit PR1 to PR5 (anxiety under queue pressure, legibility/affordance clarity, self-efficacy, facilitating conditions, and cumulative load) to explain variation in avoidance, help-seeking, recovery, and bail-out outcomes.
Overview:
Hear × Feel → avoidance/help-seeking (PR1): queue noise and time pressure elevate anxiety; visible support increases the likelihood that stalls convert to completion.
See → Do → error loops (PR2): poor legibility and ambiguous affordances increase extraneous load and repeated tapping/restart behaviour.
Cumulative load at payment (PR5): legibility burden combined with multi-step payment uncertainty increases bail-out to counter; simplifying flow reduces this risk.
Table 4 presents the coding hierarchy from first-order codes to axial categories and aggregate themes, with each theme aligned to its dominant mechanism pathway: PR1 (queue-pressure anxiety), PR2 (presentation quality and extraneous load), PR3 (self-efficacy and persistence), PR4 (facilitating conditions and recovery), and PR5 (cumulative load at payment). Findings are based on full-sample pattern familiarisation (n = 30), with illustrative mechanism-focused quotations and deep case interpretation drawn from the purposive subsample (n = 15).
Before presenting the themes, we clarify what empathy mapping adds beyond a conventional usability synthesis. Usability studies often stop at “what went wrong” (small text, confusing steps). Empathy mapping lets us trace how the breakdown unfolds in context by linking quadrant cues into short mechanism chains, for example: what participants hear and see under queue pressure shapes what they feel, which then shapes what they do (persist, seek help, or bail out). This is why we report theme evidence in a cue → emotion/cognition → coping action → outcome structure. The theoretical value is that it distinguishes bail-out from recovery as qualitatively different outcomes produced by different combinations of anxiety, self-efficacy, and facilitating conditions, rather than treating all errors as equivalent. To synthesise this mechanism-oriented reading, Table 5 presents the cue-to-outcome chains across themes, linking empathy-map evidence (See/Hear/Think/Feel/Do) to interpretation, intervention targets, and expected behavioural shifts. The following Sections elaborate on these mechanism pathways with transcript-grounded evidence and implication mapping.
Empathy-Map Interaction Patterns, Underlying Mechanisms, and Design Interventions.
Queue Pressure and Anxiety (Hear × Feel) Shape Avoidance and Help-Seeking
Queue conditions and ambient noise (Hear) interacted with worry and uncertainty (Feel) to undermine confidence at key moments of kiosk use. A large majority, 23 of 30 participants, reported difficulty registering touch inputs, often needing multiple attempts; as P01 noted, “The screen sensitivity sometimes doesn’t register my touch, I have to press twice.” The absence of immediate, trustworthy feedback, including visual, auditory, or haptic confirmation, further amplified hesitation and left users unsure whether the system had accepted an action or placed an order. Eighteen participants also described feeling rushed when people queued behind them, which compounded anxiety and shifted behaviour toward avoidance or help-seeking. As P03 stated, “I get very, very stressed up when there’s people line up behind me.” This Hear × Feel pattern is consistent with evidence that older users experience anxiety and dissatisfaction when technology does not perform as expected (Kim & Park, 2022). Participants also emphasised that visible assistance can reverse stall moments. For example, P13 reflected, “There was a staff… Press this button and that… she was very kind.” The importance of responsive, proximate support aligns with service quality evidence linking interaction quality to satisfaction and continued use of self-service technologies (Baba et al., 2023). Overall, extraneous load created by weak or delayed status cues, especially under queue pressure, nudged many users toward stepping aside or seeking help; where facilitating conditions were visible, stalls more often recovered to completion.
Building on prior usability research, this study details a process pathway linking contextual triggers to emotional responses and subsequent behavioural adaptation or withdrawal, reframing kiosk exclusion as mechanism-sensitive rather than feature-only.
Legibility and Affordance Clarity (See → Do) Reduce Extraneous Load and Error Loops
Barriers clustered around what participants could comfortably see and confidently do. Small base text, low contrast, and crowded layouts made on-screen information difficult to parse, consistent with design recommendations for enlarged text and higher contrast for older users (Na et al., 2021). Participants also reported ambiguous navigation and unstable action cues. As P05 noted, “they don’t put one, two, three, four… so… confusing.” At decision points, ambiguity in payment options further increased hesitation; P08 explained that the difficult point was “on the payment side… whether I’m using a credit card or the debit card.” These See → Do breakdowns increased extraneous cognitive load by redirecting limited resources from task progression to interface deciphering. Deep menu structures and unclear confirmation states further eroded confidence, while language-selection difficulty reduced accessibility for users less comfortable with default interface language (P12). This pattern is consistent with broader evidence that poor accessibility and convoluted interfaces suppress adoption and confidence among older adults (Zulkafly et al., 2021), and that system quality, security, and accessibility shape continued SOK use (Hashim et al., 2023). Taken together, findings support a cognitive-friendly presentation: larger base text and higher contrast, stable button placement with verb-first labels, shallow paths to core functions, and clear confirmation cues to reduce error-restart-abandonment loops.
Trust, Feedback, and Recovery at Payment (Intrinsic Load × Support): Explaining Bail-Outs Versus Recovery
As summarised in Table 5, the practical value of the empathy-map approach is not limited to identifying friction points; it shows where and how low-cost support interventions can interrupt escalation and increase recovery probability. Prior studies have already documented recurring kiosk barriers for older adults, including small text, visual clutter, unclear instructions, and payment complexity. The contribution of this study is not to claim novelty in the barriers themselves, but to explain how and when these barriers translate into observable behaviour in a public, time-pressured restaurant context. By analysing empathy-map quadrant interactions, what participants hear and see, what they think and feel, and what they say and do, this study identifies escalation pathways linking contextual cues to internal states (anxiety, uncertainty, low confidence), cognitive demands (extraneous versus intrinsic load), and coping outcomes. This yields a mechanism-based distinction between recovery (persistence supported by clear feedback and accessible help pathways) and bail-out (withdrawal when uncertainty peaks, particularly at payment and confirmation steps). Theoretically, this refines kiosk usability accounts by specifying boundary conditions for persistence: the same interface friction can lead to completion or abandonment depending on perceived social pressure and the visibility of facilitating conditions.
Payment was the inflection point where successful completions often faltered. Multi-step confirmations, ambiguous choices, and weak success signalling increased intrinsic load and left users uncertain about whether a transaction had gone through. P08 described practical breakdown at this stage: “the only part some time I get stuck is on the payment side… whether I’m using a credit card or the debit card.” P14 similarly highlighted recurring uncertainty at payment, especially after infrequent use. Participants also described moving away from the kiosk when confidence dropped. As P06 stated, “I prefer to pay the cash at the counter rather than taking a chance to make a mistake again.” These responses show that facilitating conditions, including staff in sight and visible in-flow recovery cues, can bolster self-efficacy and convert stalls into recoveries.
Managerially, the empathy-map framing clarifies that support is not merely a matter of staffing level. Support is the perceived availability of recovery at the moment of uncertainty. Two low-cost interventions are strongly indicated by participant behaviour: first, an always-visible help affordance that does not require leaving the flow; second, a predictable recovery pathway with clear back/undo, clear edit-order options, and consistent confirmation language. This reframes the practical question from whether usability issues exist to which recovery cues prevent abandonment under pressure, and at which step those cues matter most. These implications remain aligned with evidence on facilitating conditions and adoption among older users (Hashim et al., 2023; Wu et al., 2023; Zulkafly et al., 2021). To consolidate the cross-theme interpretation, Table 5 summarises empathy-map pathways by linking quadrant cues (See/Hear/Think/Feel/Do) to behavioural outcomes, intervention targets, and expected shifts in user response. All propositions (PR1–PR5) informed interpretation, with stronger empirical saturation in PR1, PR2, and PR5 and comparatively weaker but relevant evidence for PR3 and PR4. A key pattern is that similar interface frictions can lead to different outcomes (recovery, help-seeking, or bail-out) depending on anxiety, confidence, and facilitating conditions.
Conclusion
This study contributes by specifying mechanism pathways in older adults’ SOK use, rather than merely restating known usability issues. Using empathy mapping with thematic analysis, we show how contextual cues such as queue pressure and social exposure interact with anxiety, self-efficacy, and facilitating conditions to shape whether users recover, seek help, or abandon the transaction. The most consequential failures occurred when extraneous interface burden and intrinsic payment complexity accumulated in the same interaction episode (Kim & Park, 2022; Nam et al., 2023; Wu et al., 2023).
The practical implication is precise: inclusive kiosk design should prioritise recovery visibility, not only speed. Clear step progression, stronger confirmation states, larger readable text, stable navigation cues, and in-flow assistance are not cosmetic adjustments. They are mechanisms that reduce cognitive burden, stabilise confidence, and improve completion among older users in real restaurant settings (Hashim et al., 2023; Leite et al., 2022; Na et al., 2021; Yaacob et al., 2022).
The study is bounded to Malaysian quick-service restaurant settings and a qualitative interpretive design. Because deep interpretive analysis focused on a purposive 15-case subsample, the study emphasises mechanism explanation and transferability over statistical representativeness. Future work should test transferability across service contexts, language settings, and kiosk platforms, and evaluate whether the proposed recovery-oriented interventions produce measurable reductions in bail-out rates under natural queue conditions.
Footnotes
Ethical Considerations
This study received ethical approval from the Research Ethics Committee of Multimedia University, Malaysia (EA0152024), on May 21, 2024.
Author Contributions
Kin Meng Cheng: Conceptualisation; Methodology; Data curation; Investigation; Formal analysis; Coding; Validation; Writing—original draft; Writing—review & editing; Visualisation. Ah Choo Koo: Validation; Supervision; Interpretation of findings; Writing—review & editing. Siew Chein Teo: Conceptualisation; Funding acquisition; Project administration; Supervision; Writing—review & editing.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Multimedia University [MMUI/230093].
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data Availability Statement
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.*