Planned serendipity: exploring tourists’ on-site information behaviour

ABSTRACT The paper investigates tourist information behaviour on-site in light of the ubiquitous access to information afforded by smartphones. The study problematizes existing literature on information search behaviour and connects it with theories of unplanned behaviour. Through the concept of planned serendipity, the paper highlights how tourist information behaviour is a complex phenomenon that does not necessarily answer to a dichotomy between spontaneity and planning. The study employs a combination of Experience Sampling Method and semi-structured interviews, which allow the researcher to collect data both during and after the trip. Four themes are identified that challenge key assumptions in tourist information behaviour literature: (1) emergent and contingent plans, (2) cognitive effort on-site: iterative and specific search process, (3) tourist-centric orientation in time and space, (4) aiming for optimization. The paper concludes that tourists’ ‘phygital’ information environment both enables and constrains serendipity. The dichotomy of structure vs. serendipity is not suﬃcient to explain the role of serendipity in travel planning, where planned and unplanned behaviour coexist. The paper also contributes to the methodological landscape of research on tourist experiences and shows that everyday technologies like the smartphone can be harnessed to access the on-site stage, reducing recall bias and increasing ecological validity.


Introduction
Serendipitous encounters on-site have long been considered an important aspect of the tourist experience (Cary, 2004;Chandralal & Valenzuela, 2013).The role of serendipity in information behaviour, although under-researched in tourism, has become increasingly important since smartphones allow tourists to carry out a significant part of their information search on-site.Recent research on tourism information behaviour has found that, given the possibility to find enough information at the destination, tourists may prefer to keep their plans flexible (Kang et al., 2020;Hwang & Fesenmaier, 2011;Wang, Xiang, & Fesenmaier, 2016).Such access to information is enabled by smartphones, thanks in particular to their ability to access the internet and its computing capabilities, as well as the location-based services they are able to offer through global positioning systems (GPS) (Benckendorff et al., 2019).The reduction or elimination of roaming fees (for example within the EU) has made online information even more broadly accessible during travel (Zillinger et al., 2018).In light of this, the information environment of contemporary tourists has been defined 'phygital', a neologism that indicates the convergence of physical and digital information (Gretzel et al., 2019).
However, the flexibility afforded by smartphones does not always lead to a more spontaneous behaviour (Kang et al., 2020;Yu et al., 2018;Kang & Lee, 2022;Vaez et al., 2020).Yu et al. (2018) reported that family vacationers felt 'spontaneous without a sense of serendipity' due to the constant use of smartphones for information search during vacation.Tourism theories of information behaviour have generally considered tourist information search separately from unplanned behaviour, but recent studies have called for the inclusion of unplanned behaviour in the theoretical understandings of travel behaviour (Hwang & Fesenmaier, 2011;Hyde, 2004;Kang et al., 2020).Zarezadeh et al. (2019) note that old models of information behaviour keep being used without revision, while other authors suggest that existing models need to be re-evaluated in light of smartphone use (Kang et al., 2020).In fact, smartphones perform different and various functions during the trip.Tussyadiah and Wang (2016) observed that the smartphone helps tourists with information search, problem solving, communication and entertainment.Moreover, they enable socialization through phone calls, text messages and social media.Smartphones offer tourists context awareness through real-time and location-based information, updates, tracking and tagging as well as information about attractions and destinations, timetables and schedules, currency conversion, QR codes and virtual guides (Benckendorff et al., 2019;Dickinson et al., 2014;Wang et al., 2014;Yu et al., 2018).
Moreover, tourist information search behaviour literature gives scarce attention to information that is not searched but encountered, for example through push recommendation systems, which is typical of the younger generations.Schultz et al. (2019) argue that millennials have an 'information will find me' attitude.To indicate that the active search component is not always present, Wilson's (2000) terminology is adopted in the present paper: instead of referring to the theoretical framework as information search behaviour, it will be referred to as information behaviour.
Serendipity, generally defined as a fortunate but unplanned discovery, is a useful concept to explore the unexpected and unplanned dimensions of the tourist experience.In information behaviour theory, the term indicates the 'chance finding of pertinent information, either when not looking for anything in particular or when looking for information on something else' (Agarwal, 2015, p. 1).The term is particularly appropriate because it refers to both a context and a behaviour, while terms like spontaneity and flexibility only refer to a behaviour or a preference of the tourist.However, the term holds a positive connotation which may not fully represent the phenomenon at hand.While previous research has positioned serendipity as opposite to planning (Huang et al., 2014(Huang et al., , 2016)), the present study suggests that the juxtaposition between planning and spontaneity is not sufficient.Tourists want to be flexible during their trip, but they also reduce the uncertainty of travel by increasing their information search on-the-go (Wang et al., 2016;Yu et al., 2018).Pointing to the coexistence of planning and serendipity, this paper adopts the term 'planned serendipity', an oxymoron that hints at the complex nature of information behaviour on-site (Mieli & Zillinger, 2020;Rennella & Walton, 2004).
The present paper aims to explore tourists' on-site information behaviour through an empirical investigation.The paper focuses on a generation of tourist that has grown up alongside the development of digital and mobile technologies, the so-called millennials or Generation Y.Although generational distinctions do not always offer a solid division between demographic cohorts, age groups are still one of the most common and useful ways to classify a population, together with the common events that shape people's lives (Li et al., 2013).This generation has been found to prefer authentic experiences, cultural exchange, and new destinations outside of the mainstream, while at the same time seeking relax, fun and escape from everyday life (Li et al., 2013;Rita et al., 2019).Moreover, tourists within this generation usually organize their trips on their own, instead of choosing packaged trips or travel agencies (Aceron et al., 2018).
The abductive and qualitative approach of the study allows to explore tourists' behaviour and challenge key assumptions underlying seminal theories such as Vogt and Fesenmaier's (1998) model of information needs, Fodness and Murray's (1999) model of information search and Jeng and Fesenmaier's (2002) model of travel planning.In order to gain access to the on-site stage of the experience, and not only to the participants' recollection of it, a methodology was developed for this study which employed a combination of Experience Sampling Method and semi-structured interviews.The development of a qualitative methodology for this investigation answers to existing calls for new approaches and a more holistic view of information search (Liu et al., 2022;Gretzel et al., 2019;Höpken et al., 2019) and aims to enrich the current landscape of research on the subject, which is largely based on quantitative methods (Law et al., 2018).
In the following sections, relevant literature on tourism information behaviour, unplanned behaviour and serendipity is reviewed.A description of the methodology follows, where the use of the experience sampling method in combination with interviews is explained.The analysis of the data is presented in the subsequent chapter, where four characteristics of on-site information behaviour are identified.A conclusions and discussion chapter close the paper with some reflections on the results and implications for future research.

Tourist information behaviour
At the core of tourist information behaviour research lies a handful of models and theories that generally date to pre-smartphone times, and at times even to pre-internet times, and cannot accurately describe the reality of tourists' 'phygital' information environment (Gretzel et al., 2019).For example, Zarezadeh et al. (2019) criticize the citation practices of Fodness and Murray's (1999) model, noting how their data was collected in the 1990s.Despite the dramatic changes in information sources following the development of the internet, social media and mobile technologies, subsequent research cited the model without ever attempting to update it, nor to overcome its methodological limitations (Zarezadeh et al., 2019).
A large part of the literature in the field of travel information search behaviour has focused on information needs, and particularly functional ones, as the main drivers of information search (Vogt & Fesenmaier, 1998;Xiang & Fesenmaier, 2020).In Vogt and Fesenmaier's (1998) expanded model, while travel information is mainly collected and used for functional reasons, five types of needs exist: functional needs (product knowledge, reducing uncertainty, maximizing utility or value and efficiency), hedonic needs (phenomenology, experiential, sensory and emotional), innovation needs (novelty seeking, variety seeking, creativity), aesthetic needs (imagery and fantasizing), sign needs (symbolic expression and social interaction) (Vogt & Fesenmaier, 1998).Choe, Fesenmaier, and Vogt (2017) reassessed the 1998 model and concluded that, while functional needs are still the main reason for information searches, hedonic, innovation, experiential and sign needs are becoming increasingly important in the post-internet era.However, by only considering information needs, these studies imply that the search for information is a strictly functional activity, as it is only carried out to satisfy needs.Other studies considered different determinants of information behaviour.For example, Cho and Jang (2008), investigated the value structure of information for vacation travellers, assuming that tourist information represents a source of value.Korneliussen (2014) added that tourism information search is a 'Do-It-Yourself (DIY) tool' to create experience value.
Seminal literature in the field claimed that tourists' choices of information sources represented distinct strategies, categorized based on three dimensions: spatial, temporal and operational (Fodness & Murray, 1998, 1999).Fodness and Murray (1997) also examined possible ways to segment tourists based on their information search strategies, based on time spent planning and number of sources used.Jeng and Fesenmaier (2002) proposed that decision-making is a hierarchical, continuous, adaptive process, which relies on three different types of decisions, corresponding to three types of information: core, secondary, and peripheral.While core decisions such as choice of destination are made early in the planning process and are rather rigid; secondary and peripheral decisions such as activities or places to eat, can be made later and are more flexible (Jeng & Fesenmaier, 2002;Kang et al., 2020).Kang et al. (2020) claim that the possibility to use smartphones for information search during the trip has the potential to change which decisions are taken before and during the trip, and therefore information behaviour should be understood beyond the three phases of pre-, during and posttrip.Hwang and Fesenmaier (2011) challenge the assumption that travellers decide an itinerary prior to their trip and follow it more or less to the letter.Wang et al. (2016, p. 6), moreover, found that due to the use of smartphones, tourists planned less or planned differently, and that they perceived 'more flexibility during trips', 'more en-route planning', 'less prior planning', and even 'more trips'.More recently, Liu et al. (2022) found that tourists use their smartphones to understand and interact with different physical and social contexts, which often result in unplanned behaviour, or in new plans.The authors found that even though plans are made before the trip, due to the use of smartphones for on-site decision-making, they can be revised and cancelled.
Studies on unplanned behaviour have attempted to explain under which conditions such behaviour occurs, and concluded, for example, that a revision of the travel plan en-route occurs in the following three conditions: new information is found on the way; a discrepancy exists between expectations and reality during travel; unanticipated constraints occur (Hwang & Fesenmaier, 2011;Stewart & Vogt, 1999).On-site decisions are believed to be light-hearted, free-spirited, hedonistic, unreflective, immediate, spontaneous, and that they do not require intensive information processing; in contrast with pre-trip decisions, defined as deliberate, purposeful and reasoned (Kang et al., 2020;Choi et al., 2012;Hyde, 2004;Hwang & Fesenmaier, 2011).Thanks to mobile technologies and constant access to information online, tourists evaluate and change their plans continually (Kang et al., 2020;Liu et al., 2022).Moreover, hedonic aspects of vacationing may lead travellers to not plan their trip in advance for the very purpose of enjoying unexpected situations during their trips (Kah & Lee, 2014).
Research also suggests that the technologies themselves with their affordances determine changes in tourists' behaviour.Kah and Lee (2014), for example, found that travellers who used global positioning services (GPS) were significantly more likely to engage in unplanned travel activities due to the fact that they could obtain information during their trip.In the same way, smartphones afford tourists ubiquitous access to information and thus have the potential to encourage more spontaneous behaviour, both because information can be accessed at any time, and because of the personalized, real-time, location-based content as well as context-aware recommendations that smartphone apps offer (Wozniak et al., 2017).

Serendipity in the tourist experience
Acknowledged in information behaviour theory but not in tourism information search literature, is the concept of serendipity.Serendipity in information search can be defined as 'chance finding of pertinent information, either when not looking for anything in particular or when looking for information on something else' and 'often drawing a reaction of happiness, surprise or simply an ahah! moment (and, sometimes, disappointment as well)' (Agarwal, 2015, p. 1).Serendipity plays an important role in how people discover, explore and learn (Björneborn, 2017).Foster and Ford (2003) propose an interpretation of serendipity as both purposive and non-purposive: serendipity can yield unexpected results from new information that is encountered on the way, but that is somehow the result of a purposive behaviour of the person, who expects to find information on the way but who does not know how, when and what they will find.By this interpretation, the person is not an information seeker, as they do not actively seek that information, but they encounter it serendipitously.
In tourism research, serendipity has been recognized as an important part of the tourist experience and one of the antecedents of memorable experiences (Andrades & Dimanche, 2014; Cary, 2004;Chandralal & Valenzuela, 2013).Huang et al. (2014) investigated independent travellers' preference between serendipity and structure.They classified independent travellers as either 'serendipitous' or 'structured', two opposite dimensions of independent travel, and found that serendipitous travellers spent less time planning their trips and searching for information, while structured travellers spent more time on travel planning and searched for more specific information on their destination (Huang et al., 2014).Serendipitous travellers, according to Huang et al. (2014), make more decisions on-site.
A related concept that is established in tourism literature is that of 'venturesomeness', that is the main characteristic of the allocentric type of tourists theorized by Plog (1974Plog ( , 2001) ) in his model of psychographics in tourism.Like Huang et al. (2014), Plog (1974Plog ( , 2001) also creates a spectrum with the two extremes of psychocentric and allocentric.The first denotes a type of tourist who privileges safe and low-risk choices, a conservative tourist who tends to plan more and choose places they already know, who prioritizes comfort.The latter is a more adventurous, risk-taking tourist who would rather explore new places and seek spontaneous experiences.Mieli and Zillinger (2020) observed that due to the new information environment afforded by mobile technologies, also termed phygital in previous research (Gretzel et al., 2019;Mieli, 2022a), spontaneity and planning are not necessarily juxtaposed but they can coexist in the travel experience.The authors proposed the concept of 'planned serendipity' as an alternative to the juxtaposition of serendipity and planning (Mieli & Zillinger, 2020).Planned serendipity is a contradiction in terms that indicates that 'the planning process can leave space for elements of surprise or novelty' and 'serendipity does not have to be juxtaposed to planning, but it can be an important element of it' .While the term serendipity holds a positive connotation, the term planned serendipity does not directly imply a judgement evaluation, and hints at how the consequences of such behaviour can be both positive and negative, or simply different than what previously thought about tourist information behaviour.
The expression 'planned serendipity' appears in management literature, and it was made popular by a book by Thor Muller and Becker (2012), titled Get Lucky: How to Put Planned Serendipity to Work for You and Your Business.However, in Muller and Becker's (2012) book the concept indicates a business strategy that encourages companies to harness unpredictable elements to their economic advantage.In the present paper, planned serendipity is used to indicate tourists' information behaviour, and is not connected with business strategy or economic gain.The term is however not common in academic literature either in the management or the tourism field.In travel-related literature, the term planned serendipity only appears in a 2004 article by Rennella and Walton, titled Planned Serendipity: American Travellers and the Transatlantic Voyage in the Nineteenth and Twentieth Centuries.Rennella and Walton (2004, p. 371), however, do not explain what they mean with the expression they use in the title and only mention serendipity once in the article, stating that 'Serendipitous encounters during their travels became an essential part of voyagers' personal, professional, and intellectual development.'.Although they point to its importance in tourism, the authors do not connect planned serendipity with information behaviour and leave space for a clearer definition of the concept.

Methodology
The study was approached qualitatively, with a three-step methodology that included: a short questionnaire to gather demographic data and information about the trip; Experience Sampling Method (ESM); and a follow-up interview.Fifteen people participated in the research (See Table 1

below).
Data was collected between May 2019 and March 2020.Due to the global COVID-19 pandemic, data collection had to be interrupted in March 2020.However, for an ESM study it is rather common to have fewer than 20 respondents (van Berkel et al., 2017).Given the richness of the data collected through the three steps, this was considered a sufficient amount of data to process, and it allowed the researcher to draw trustworthy conclusions, although the research design neither allows nor aims for generalization.
Research participants were selected according to two main criteria: age and education level, which are closely associated with information behaviour (Coromina & Camprubí, 2016;Kim, Xiang, & Fesenmaier, 2015).Participants were born between 1980 and 2000, a generation that experienced travel both with and without smartphones and ubiquitous internet connectivity, and therefore can evaluate their impacts on their experience somewhat more critically, or at least be more aware of them.Research on tourists' preferences and typologies has attempted to apply models such as Plog's to millennials, without however being able to clearly identify a distinction between psychocentric and allocentric tendencies in millennial tourists (Aceron et al., 2018).Tourists belonging to this generation have characteristics that could be attributed to both extremes of the spectrum and would therefore generally fall within the middle of the curve.In fact, research has shown that millennials generally prefer authentic experiences and seek cultural exchange, being comfortable with new destinations (Li et al., 2013).Moreover, they show a preference for challenge and adventure but appreciate other, more relaxing experiences as well (Rita et al., 2019).
All participants had attended tertiary education, as this group has been found to spend more time on trip planning than others (Coromina & Camprubí, 2016).Use of smartphones during travel was also included as a selection criterion in the research invitation.Although psychographic data was not collected for the present research, the demographic and trip-related data collected suggests that most respondents would fall within the middle-allocentric side of the spectrum, in line with previous research.Except for one respondent who was on a chartered trip, all others had planned the trip on their own.All of the respondents planned daily activities on their own, which is consistent with Plog's (2001) allocentric type.Only three of the respondents had chosen places they had already visited, two of whom were visiting friends or relatives.The remaining 12 participants chose destinations they had not already visited, showing characteristics of the allocentric type of tourist (Plog, 2001;Litvin, 2006).Regarding activities on-site, respondents chose a variety of activities, including visiting museums, eating local food, spending time at the beach or by the hotel swimming pool, taking part in local experiences such as yoga retreats and cinema, and many more.Therefore, categorizing respondents based on their activities on-site would not be feasible.In Table 1 below, demographics and trip-characteristics are shown, including destination, travel purpose and type, and trip design.As shown in the table, the sample is quite heterogeneous in terms of nationality, length of trip and distance travelled but with a prevalence of urban tourism and self-planned trip.
Participants were recruited on a voluntary basis.Fictional names were used to anonymize interviewees, following general best practices in social research (Flick, 2018).No compensation was offered to participants.While compensation can be an effective way to encourage participants to join the research project, Scollon et al. (2003) found that it could result in a poor quality of the data due to lack of intrinsic motivation.
Interest from prospective participants was collected through an online form, where information was collected about their demographics and the trip they had planned.The second step consisted of gathering data during the trip through the Experience Sampling Method, which is an established method in clinical psychology, introduced in the late 1970s by Csikszentmihalyi et al. (1977).The method has so far seen scarce application in tourism, perhaps due to the twofold challenge of requiring active participation by the respondents during their trips and the necessity to provide them with research instruments to administer the surveys.In fact, until smartphones became widely available as personal technologies, researchers needed to provide beepers, booklets or recorders to participants (Hektner et al., 2007).The method can now be conducted on participants' own smartphones; however, the researcher still needs to provide the software (van Berkel et al., 2017).Some examples of Experience sampling method in tourism are Birenboim's (2016) and Shoval, Schvimer, and Tamir (2018) quantitative studies of tourist experiences in time and space, as well as Quinlan Cutler et al. (2018), who discuss how they used the method in qualitative tourist experience research.
With ESM, participants receive daily mini-questionnaires to fill in as soon as possible and the questionnaires focus on participants' momentary behaviours and feelings (Hektner et al., 2007;van Berkel et al., 2017).Participants downloaded a software application on their phone, free of charge, and received a notification at random times within a set time range, once or twice per day, for 4-7 days (depending on the length of the trip).Upon receiving the notification, participants filled in a short questionnaire, which took about two minutes to complete and contained different types of questions, including: multiple choice, scales, open questions (cf.Mieli, 2022b).The questionnaires were built using branching logic, where certain questions depend on previous answers (for example, if the answer to 'Did you make plans for today?' was No, then the question 'Have your plans changed?' would not appear).Therefore, not all respondents answered the same questions, the same number of times.The purpose of this method was to access tourists during the trip, collecting data at a point in time as close as possible to the actual events reported.Moreover, smartphone use involves a number of actions that users make without realizing it or without giving it too much importance, which would most likely be forgotten until after the trip, when data would be otherwise collected.The response rate among participants was very high, with respondents answering all or almost all the questionnaires they received.Table 2 below shows how many questionnaires each respondent received and how many they answered throughout the trip.
The third step was a qualitative, semi-structured interview held in person or over the phone after the trip.Qualitative interviews allow for a greater depth and richness in the data collected (May, 2011).The duration of the interviews ranged between 20 min and 1 h.ESM data was used as a starting point for questions, and interviewees were often asked to explain or expand on the answers given in the questionnaires.All interviews were recorded and later transcribed manually by the researcher.Fifteen people completed all the steps.In one case, the experience sampling data was lost due the app malfunctioning, but the participant (Ines) agreed to be interviewed anyway.Given the richness of the interview, this was included in the dataset.The whole dataset, therefore, consists of 14 sets of surveys and 15 interviews.Since the ESM surveys were answered several times by each respondent over the course of their trip, the total amount of surveys collected is 93.
This methodology allows for a deep investigation into the behaviour of tourists as well as their recollection of the experience.While these methods do not allow to conduct any statistical analysis that may indicate causality or correlation, ESM has the advantage of a high ecological validity, as it can capture respondents' 'experience before it has a chance to be filtered by memory or altered through subsequent self-reflection' (Hektner et al., 2007, p. 110).In fact, questions are immediate and refer to events that are occurring or have just occurred.Moreover, randomized signals ensure reduced reflexivity bias, and the relatively high number of signals allows participants to get used to recording their behaviour (Hektner et al., 2007;Mieli, 2022b).
A thematic analysis was conducted on the dataset, combining survey data and interview transcripts, following the general principles of grounded theory (Bryant & Charmaz, 2007).Due to the small sample, ESM data was not analyzed statistically, but it was summarized and used as input for the interviews.In the analysis, the concept of planned serendipity was used as a sensitizing concept, that is a tentative conceptual category used in the analysis as the basis for theory building (Bryant & Charmaz, 2007).Interesting points were coded through a first round of open coding on both ESM data and interviews.The first cycle of coding resulted in a codebook including over 80 codes.A second round of coding identified ten broader codes, for example 'optimization', 'tension between knowing/not knowing', 'change of plans', etc.A third round of coding focused on the set of ten codes, and the data was then further elaborated into four themes, which will be presented in the next section.

Thematic analysis
The analysis of ESM data and interview transcripts resulted in four themes: emergent and contingent plans; cognitive effort on-site; tourist-centric orientation in time and space; aiming for optimization.Through the four themes, the paper shows how tourists' phygital information environment both enables and constrains serendipity, producing the phenomenon called here planned serendipity.

Emergent and contingent plans
Most interviewees defined their initial travel plan as an idea, a rough plan.They had lists they had prepared in advance, but these lists were often not actually used to plan each day, were left at home, or used retrospectively to check things off.When using the list, respondents mostly checked online for more specific information.Most commonly, respondents planned the 'what' but not the 'when'.For example, Alma made a list of things to see, a 'general idea', and once at their destination, she would decide which things from the list to see when, depending on where she was and how far the attraction was.When asked in the ESM questionnaires whether their plans for the day had changed, in most cases plans were flexible, had changed or there was no plan at all.In most cases a detailed plan for each day or itinerary was only decided the night before or in the morning.For example, Fritz explains: 'we knew where we wanted to go and we booked some of the accommodation and some of the ferries, so that gave our entire trip some kind of schedule.
But usually, we used the evenings or early morning when we knew about the weather'.In fact, unpredictable factors like weather and physiological factors like hunger and tiredness were often mentioned to show how the smartphone allows for a more flexible trip.These data highlight the importance and extent of on-site planning and information search for smartphone users.As Kah and Lee (2014) suggested that having a device that allows for gathering information on-site has a strong influence on tourists' behaviour.Data from the present research suggests that smartphones have the same effect on travel plans as Kah and Lee (2014) observed in relation to navigation systems.While these data are in line with research that views travel plans as dynamic and evolving throughout the trip, they contradict Hwang and Fesenmaier's (2011) and Stewart and Vogt's (1999) claim that specific conditions must occur during the trip for unplanned behaviour.Instead, not planning was a deliberate choice, based on the knowledge that information would be easily accessible during the trip.
Plans are not only made anew, revised, cancelled during the trip as found by Liu et al. (2022), but they are purposefully made flexible to make the most of on-site information search.Even plans made on-site are not too detailed or specific, they are continuously revised, cancelled and remade based on information encountered on-site.It is hard to distinguish between what is planned and what is unplanned: it is not necessary to make detailed plans in advance because tourists know they will have access to information at any time, and their needs for information are thus postponed.The analysis points to how travel plans are contingent and emergent, and how the availability of information on-site through smartphones makes the information search dynamic.Tourists' information behaviour cannot be directly ascribed to specific needs or strategies, contrarily to existing literature (Vogt & Fesenmaier, 1998;Xiang & Fesenmaier, 2020).Additionally, respondents' behaviour was not constant throughout the trip: the ESM answers show that each participant's reliance on their smartphone for information search varied from day to day.This suggests that creating categories and types of tourists based on information strategies and use of sources, like Fodness andMurray's (1998, 1999) model, may conceal the complexity of tourists' actual behaviour on-site.

Cognitive effort on-site: iterative and specific searches
Participants often described their information search on their smartphone as iterative.A Google search was often the point of departure and searching directly in Google maps was also common.Most respondents were using up to 2-3 web pages each time, going back and forth or opening more than one page at once, reading or skimming through several web pages and then narrowing down the search based on proximity and interests.Liisa explained her process: 'I wrote in the source field 'madeira jeep tours' then I ended up on TripAdvisor, check few of them, check what people had been writing, and then I also check some reviews.[…] Sometimes the information on the website was good, but sometimes it wasn't enough, so you went and checked what the others have been writing.Then redefine the search a couple of times'.Further, respondents described their search as 'constant' (Gabriela), 'short but efficient' (Fritz) or, as Olof said: 'I am very quick to pick [the phone] up out of my pocket to check even minute things.For example, directions I check over and over and over and again.Like you walk a little bit, you pick up the phone and you check the direction again'.
This iterative characteristic of information search shows how tourist information behaviour is a dynamic process.Information is sought at all stages of the trip but is never fixed: new information is searched and collected throughout the trip, decisions are made and remade.As highlighted by previous research, the use of the internet causes decisions and information needs to be postponed (Xiang et al., 2015;Mieli & Zillinger, 2020).
Besides being iterative, information search on the phone was also described as specific.A phone search for information differs from using other sources (guides, books, brochures, information centres) in that the person needs to look specifically for something, they need to know what they are searching in order to enter a query string in the search field when using a search engine.This makes information search more intentional and specific, narrowed to what the tourists already know they should be looking for.In this sense, information is not encountered casually or serendipitously, but specifically sought out.This kind of continuous, efficient, specific search requires competence and processing of large amounts of information.Fritz, for example, explained: 'I have been a student for five years.I'm very good at cross reading and […] collecting information with as little effort as necessary'.This contradicts previous research which states that decisions made during the trip are mostly light-hearted, free-spirited, hedonistic, unreflective, immediate, spontaneous and do not require intensive information processing (Kang et al., 2020;Choi et al., 2012;Hwang & Fesenmaier, 2011).
The iterative and specific character of information search casts doubts on the assumption that information search is carried out in stages, and that different types of information are sought in different stages of the trip, with less cognitive effort in the on-site stage (Choi et al., 2012;Jeng & Fesenmaier, 2002;Hyde, 2004;Hwang & Fesenmaier, 2011).The specificity of the information search also has important consequences for the serendipitous element of information encounters: it reduces the possibility to find pertinent information by chance, and those 'ahah!Moments' (Agarwal, 2015, p. 1).

Tourist-centric orientation in time and space
Several participants used expressions such as 'ok, I'm here, what is around me' to describe how they used online maps to understand where they were located and what was available nearby.The process consists of assessing the current situation, determining what information is needed, collecting such information online, then using maps to navigate.Orientation is not done in space only, but also in time, because people search for what to do next, based on where they are at any given time.For example, Gabriela described: 'it would be like: OK let's go for lunch, I'm going to check where it is.At the same time, OK, I know where it is, but I have to use my Google maps to go there so I'm holding [the phone] the whole time.Then we're there and we need to look for where the beach is.OK, I'm going to find where the beach is, OK, let's go to this one, OK I'm going to use my Google maps to navigate'.
In another case, Alma was planning to visit a monastery, which was on her initial list of things to do.However, when she was in the city centre, she was looking for information on her phone and found that there was a zoo nearby, so she decided to take her daughters to the zoo instead of the monastery because it was closer.This instance shows how the choice between two completely different attractions was made based on proximity and immediacy.These two elements are therefore important factors in decision-making, as both the search process and the decision are rather immediate and based largely on what is near in space and time.This expands on Xiang et al.'s (2015) findings that decisions are postponed in connection with smartphone use and on Liu et al.'s (2022) statement that smartphones are used by tourists to interpret, assess and comprehend their immediate physical contexts.
Orientation, on the other hand, is not only the ability to see what is around, but also to know where we are and where we are going.Some respondents complained that with the smartphone and digital maps, they do not bother to learn where they are, where landmarks, attractions and reference points are, nor to memories itineraries and names of places.Kaisa complained that before smartphones she would always get a paper map and learn to orient herself in the destination: 'nowadays I can't do that anymore.I don't know if it's because I don't use enough time for that or I'm just stupider or my brain doesn't function the same or just that I'm so used to being with my phone that I don't need to know these things anymore'.This aspect has been overseen by previous research on travel information behaviour, which rarely focuses on what tourists do not learn and what they are not able to do in connection with the use of the smartphone.
The assumption critiqued here is that the object of the information search is the destination itself, while it appears from these results that the object and centre of the information search is the tourist.Moreover, while it is generally assumed that the aim of information search is gaining knowledge, and research has shown that the tourist experience has epistemic value (Williams & Soutar, 2000), these findings show that the epistemic dimension of information search are different when information can be retrieved at all times and on-the-go.

Aiming for optimization
One aim of the on-site search appears to be the optimization of the trip, that is, tourists try to make the most out of their limited resources, especially time and energy.Through the smartphone, and in particular thanks to constant connectivity, online maps, location-based services and specific apps, tourists can avoid many of the negative experiences that are common in travel: getting lost, staying hungry longer than necessary, getting caught in bad weather, getting stuck in traffic, spending unnecessary money or being taken advantage of.
Participants believed that such behaviour could either foster or hinder serendipity, depending on one's perspective on it.While some thought that this made them more spontaneous because they could 'spontaneously check' everything (Fritz, Kaisa); others believed it made them less spontaneous because they ended up 'checking and double checking' everything (Gabriela, Olof).For instance, Olof explains how making the plan optimal can end up hindering serendipity: 'if google maps says 18 min with the car to there, then you plan 20 min but if you have no idea, you may plan 40 min and then you have the time!And if you plan 40 min and there is not so much traffic, you are like oh here is something, let's stop here for a bit, we have some time … then you live more in the moment'.
These data highlight the connection between optimization and serendipity, which are not necessarily mutually exclusive, but certainly influence each other.Although the aim of information search has always been believed to be to improve the experience, reduce risk and uncertainty and satisfy those needs identified by Vogt and Fesenmaier (1998), the findings suggest another view, in which the aim of information search is to optimize the experience through the computational capabilities of the phone and the location-based services it offers.

Discussion and conclusions
The study has explored the use of smartphones for information search on-site and identified four themes that highlight how tourist behaviour has changed in light of smartphone use.The contribution of the paper is both theoretical and methodological.Methodologically, the paper contributes to the field by applying an innovative method such as ESM in a qualitative study.Such a method not only allows the researcher to gather data on-site and virtually follow the participants for several days while they travel, but it does so in a relatively simple and cost-effective way, using participants' own smartphones as a research tool.The results have shown that comparing and combining ESM data with follow-up interviews can yield interesting results and a more accurate view of tourists' actual behaviour while they travel, increasing the ecological validity of the study, which reinforces the trustworthiness of the qualitative methodology.Despite the scarce use of the methodology in tourism research, ESM has great potential for investigating on-site experiences unobtrusively and easily.
Theoretically, the paper contributes to the field of tourists' information behaviour by exploring the concept of planned serendipity and illustrating how planned and unplanned information behaviour coexist in the tourist experience.Research into unplanned behaviour has shown the importance of on-site decision-making, which can stem from a general preference for flexibility and the wish to base decisions on information encountered at the destination (Hwang & Fesenmaier, 2011;Kah & Lee, 2014;Kang et al., 2020;Wozniak et al., 2017).The idea of a plan made before the trip and followed to the letter has been challenged, because tourists now change and evaluate their plans continually (Kang et al., 2020;Liu et al., 2022).
The present paper addresses this issue by proposing the concept of planned serendipity as a way of pinpointing what happens when tourists can continually change their plans based on the situations they encounter on-site.Such behaviour is supported by the smartphone, which allows tourists to access up to date information at any time and anywhere.The paper offers an empirically grounded critique to in information search behaviour literature.The characteristics of information search identified through the four themes show that existing assumptions in tourist information theory must be questioned in light of smartphone use at the destination, and form the basis for four counter claims, summarized in Table 3.
The assumption that tourists have information needs and distinct strategies for their information search is challenged and an alternative view is proposed, where information search is more contingent and emergent.The notion that information search is carried out in stages and on-site information search only regards simpler choices without high cognitive effort, can be questioned with a continuous and progressive view of information behaviour.Moreover, if the object of the information search is not the destination but the tourist, and improvement of the experience is achieved by optimizing choices and resources, it is easier to understand why tourists feel more flexibility in their choices while at the same time feeling less spontaneous.In light of these reflections, the concept of planned serendipity can be a useful bridge between the apparently opposite concepts of spontaneity and planning.The dichotomy of structure vs. serendipity is not sufficient to explain the role of serendipity in travel planning, because mobile information technologies allow a degree of spontaneity even in structured plans.Spontaneity and planning are not opposites, but complement each other in a flexible travel plan, where room is intentionally left for serendipitous encounters.
The present study has sought to uncover and explore the complex relationship between planned and unplanned behaviour in relation to information on-site.With an innovative qualitative methodology, it has focused on the underlying mechanisms of such behaviour.However, further research might want to utilize this knowledge to predict tourists' behaviour as well as to develop guidelines for destinations.For example, a correlation between planned serendipity and travel motivations, or with tourists' psychographic data might yield interesting results of practical utility for tourism practitioners.
a VFF = Visiting friends and family.CURRENT ISSUES IN TOURISM