Spontaneity and planning in arts attendance: insights from qualitative interviews and the Audience Finder database

ABSTRACT Arts managers frequently use customer relationship management systems to identify early and late ticket bookers, but to date there has been no comparable investigation of spontaneity and planning through qualitative academic audience research. This paper combines two radically different datasets to draw new insights into booking patterns of audiences for contemporary arts events. Quantitative data from Audience Finder has been analysed to look for trends in early and late booking amongst audiences for contemporary art forms. Qualitative data has been drawn from the Understanding Audiences for the Contemporary Arts study, which used in-depth individual interviews to investigate the contemporary arts attendance of audience members in four UK cities. Interpretative Phenomenological Analysis was then used to draw out insights about where the purchasing point sits within the longer decision to attend. Following a review of marketing and audience research literature on the decision to attend, we present the findings from each of these analyses, looking at moments where they confirm, supplement, contradict, or say something completely outside the remit of the other dataset. We show how the timescale of the decision to attend is influenced by (1) art form conventions and price, (2) geographical region and availability of the arts, (3) attending arts events with companions, and (4) personal preference for planning or spontaneously choosing activities. We end by suggesting a new three-part model for understanding booking patterns, and considering how these insights might be acted upon by arts organisations.


Introduction
Arts organisations typically encourage audiences to book tickets early; this strategy is often key to ensuring a large audience for an event, and means that arts managers are less reliant on walk-up attenders to ensure they break even. However, as subscription ticket sales continue to decline (Hall, Binney, & Vieceli, 2016), and arts organisations are witnessing a shift towards late booking amongst audiences (McClure, 2019), arts marketers are turning to new strategies such as dynamic pricing, to incentivise early booking (Donald, 2017). In a time of precarious arts funding in the UK, examining patterns of attendance can help arts organisations make data-informed decisions about programming and ticket pricing (Foreman-Wernet & Dervin, 2017).
While the decision to attend has been explored in many previous audience development projects (McCarthy & Jinnett, 2001;Wiggins, 2004), little is known about the factors that influence early or late ticket-booking amongst arts audiences. Similarly, whilst much analysis is done of transactional data within the sector (by organisations such as The Audience Agency, TRG, Baker Richards, Purple Seven, Spektrix and individual venues), typically this analysis is not made available as part of the academic literature, nor is it sector-wide in its perspective and cross-referenced with large-scale qualitative analysis.
This paper draws on recent developments in audience research to shed light on planning and spontaneity amongst arts audiences. We investigate the booking patterns of arts audiences as evidenced in two radically different datasets: Audience Finder booker data, a national dataset of transactional booking data from venues across England; and qualitative interview data from Understanding Audiences for the Contemporary Arts (UACA) project. In doing so, we uncover the influence of access and availability, co-attendance, and personal preference for early and late booking, demonstrating the value of co-analysis of independent datasets for revealing new insights into audience behaviour.

Understanding the decision to attend
The focus of this paper is arts attendance, however as McCarthy, Brooks, Lowell, and Zakaras (2001) have noted, the term "attendance" is used in the literature to refer to a variety of domains of arts engagement. We recognise the complexities of defining consumption of the arts; terms such as visitors, audiences, consumers, participants and fans are often interchangeable in both theory and practice. In this paper, we use "attendance" to mean the act of physically attending an event or exhibition at an arts organisation, whether free or ticketed.
The decision to attend an arts event can be seen as a subset of decision-making theory, the facet of consumer behaviour studies traditionally associated with exploring the cognitive processes involved in the purchase of products (Engel, Blackwell, & Miniard, 1990). Various models have been developed to explain how internal personal factors and external social variables can influence the decision-making process: Theory of Reasoned Action (Ajzen & Fishbein, 1980); Theory of Planned Behaviour (Ajzen, 1991); AIDA (attention, interest, desire and action); five stages of decision-making (problem recognition, information search, evaluation of alternatives, purchase, post-purchase evaluation) (Brassington & Pettitt, 2013); and hierarchy of needs (Maslow, 1954). Each of these models centres on a recognition of need, an awareness of possible options, an evaluation of desirable alternatives (which is both rational and enculturated), an intention to purchase, and the actual purchase, with post-purchase evaluation feeding back into potential future purchases.
While these models can be tentatively applied to arts attendance (Walmsley, 2014), as Hirschman and Holbrook (1982) have noted, consumers tend to be treated as logical agents, whereas a more experiential and hedonic model of consumption may be more appropriate for the decision to attend. The hedonic model emphasises the ways in which consumption is driven by the symbolic value of the product and the purchasing decision is shaped by the consumer's multi-sensory, fantastical and emotional experience with the product. Boorsma (2006) and Fillis (2006) also argue that an understanding of arts consumption must include an appreciation of the aesthetic and symbolic motivators, which is difficult to interpret using traditional marketing frameworks. In addition, production and consumption differs in the arts to other products because artists create their offerings first and then present them to audiences, rather than necessarily creating what the market wants (Fillis, 2006;O'Reilly, 2005).
The experiential and aesthetic qualities of arts engagement are therefore often missing from models of the decision to attend. Even McCarthy and Jinnett's (2001) decisionmaking framework specific to arts engagement, which recognises the importance of the individual's background, beliefs, intentions and experience, fails to account for the appeal of the aesthetic experience, or to explain why an audience member would attend one event over another. Bradshaw, Kerrigan, and Holbrook (2010) argue that understanding arts consumption through an "experientialist" lens foregrounds the multiple use-values of art as an aesthetic and social experience. Nevertheless, there remains a paucity of research into how information-processing or hedonic models function in actual ticket booking decisions, how the decision to attend may differ between art forms, and the specific circumstances behind attendance at an arts event as told by audience members.
Nevertheless, where decision-making theory is helpful is in distinguishing between different stages in the decision to attend. Most notably for this paper, since the arts are temporal experiences, the decision to attend may occur in advance of, or concurrently with, the arts experience. Arts events can have limited availability and for events with high demand, tickets may be purchased a long time ahead of the event. Furthermore, the decision to attend may be temporally separated from the purchasing of a ticket; the Theory of Planned Behaviour (Ajzen, 1991) provides a useful delineation between intention and actual purchase, therefore suggesting that on-the-door sales may still be the result of a much earlier decision to attend. The mixture of quantitative ticket sales data and qualitative interview data in this paper enable us to tease out the relationship between intention, ticket purchase and attendance.
Much analysis has been done within the arts industry to identify trends in booking patterns including lead time of ticket purchase. While state-funded organisations such as The Audience Agency make some of their analyses public, the insight into lead times at commercial organisations such as TRG, Baker Richards, Purple Seven, and Spektrix, as well as that done in-house in arts organisations, is typically kept confidential. Some booking patterns are easily observable by audience members and researchers alike; for example, donors and subscribers are often given early access to tickets prior to the general release, therefore high-frequency and loyal attenders are often some of the earliest bookers. Analysis from Eventbrite (2015) suggests that free tickets on their platform tend to be booked last-minute, which supports the intuitive relationship between cheap tickets and late booking, but there is a lack of publicly-available robust analysis demonstrating this relationship. Furthermore, while ticket sales data can reveal largescale trends in booking patterns, this data cannot often place the point of purchase in the longer decision to attend.
What is absent from this literature is exploration of what factors induce a potential audience member to commit to attendance, whether that is a year in advance, or as they walk past the door. How important is the artistic offering and how important are the extrinsic factors, such as finding any event that fulfils a desire to relax or socialise? Is the experience of an arts event different depending on whether it was booked in advance or spontaneously chosen on the day? And where does the point of purchase occur within this potentially longer decision to attend? The combination of Audience Finder data and qualitative interview data offers a rare opportunity to address these questions by identifying large-scale trends amongst UK arts audiences, as well as exploring personal narratives of arts engagement from contemporary arts attenders.

Materials and methods
This paper draws on quantitative data from the Audience Finder database and qualitative interviews from the UACA study. These two datasets were collected entirely independently from one another, and therefore no direct relationship is assumed between them, except that they emerge from the same population: the UK arts-attending public. We combine these two datasets in the manner described by Miles and Sullivan (2012) in a previous issue of this journal, where qualitative and quantitative datasets are brought "into dialogue rather than 'triangulated' on the assumption of a corroborative relationship" (p. 312). This methodological eclecticism echoes the multiple research methods employed in largescale projects such as Understanding Everyday Participation (Miles & Gibson, 2016) and advocated in Mason's (2006) paper "Mixing methods in a qualitatively driven way". The analysis below asks the same question of both datasets: what is the timeline for deciding to attend an arts event? The limitations of both datasets and the extent that they can be co-analysed in this way becomes clearer through understanding the data collection methods employed.

Data collection
Understanding Audiences for the Contemporary Arts The qualitative data discussed in this paper consists of 135 semi-structured interviews with contemporary arts attenders in Bristol, Liverpool and London (45 participants in each city). This data was collected as part of the Understanding Audiences for the Contemporary Arts study, a 2.5-year research project exploring how people engage with new artworks across multiple art forms. Having begun with an exploratory pilot study in Birmingham (Gross & Pitts, 2016), the project has expanded to Bristol, Liverpool and London, working closely with arts organisations to understand their current and potential audiences. 1 Interview participants were recruited through five partner arts organisations in each city, all of whom presented work which they self-defined as "contemporary". Recruitment was carried out both online through partner organisations' social media and mailing lists, and in person by approaching audience members at arts events. Participants were identified as being at least somewhat engaged with partner organisations (by attending events, following them on social media, or subscribing to a mailing list), but we were careful to select participants who represented a range of different levels of arts engagement and demographic characteristics. The most challenging part of recruitment was in finding people who were not professionally involved in the arts, whose views might be skewed by their knowledge of the arts sector, and therefore not representative of the wider population. However, recruiting in person and partnering with organisations who attract diverse audiences ensured that our sample was large, varied and robust.
The interviews investigated how the contemporary arts fitted within the participants' overall cultural engagement, therefore the data collected covers art forms across boundaries such as contemporary and traditional, mainstream and alternative, commercial and public-funded, high-art and popular culture. The interviews lasted an average of 48 minutes and covered the following topics: arts attendance patterns, the decisionmaking process, and the value of arts attendance in their lives (c.25 minutes); the arts where they live, reflecting how the cultural offering in their city shaped their engagement (c.10 minutes); and routes to engagement and the ways in which their engagement fits with their other interests today (c.10 minutes). The interviews were recorded on a portable audio device and transcribed verbatim.

Audience Finder
The quantitative analysis in this paper is focused on the ticket sales data in Audience Finder. Analysing the whole database would have been unfeasibly time-consuming and so it was necessary to limit the scope of our analysis to ticket sales relating to these twelve genres, which still amounted to 28.5 million bookings: These twelve genres were chosen as those which were closest to the idea of "contemporary arts" investigated in the UACA project. Arts events were assigned to these genre categories by the presenting organisation; the Audience Agency's genre descriptions ensure that inconsistencies in definitions are kept to a minimum.

Analysis
UACA data was analysed thematically, using Interpretative Phenomenological Analysis (Smith & Flowers, 2009) to draw out insights about where the purchasing point sat within the longer decision to attend. Two researchers carefully read each transcript, identifying data that related to lead time, planning and spontaneity in the participants' comments. Themes and codes were then compared and verified. The selected interview quotes were further analysed to categorise into four emergent subthemes: differences in booking patterns for different art forms, differences across the three cities, the influence of arts companions, and personal preference for planned or spontaneous attendance. Care was taken to identify both anomalies and common trends in the data, especially where those trends cut across geo-demographic characteristics or differences in arts engagement.
Meanwhile, Audience Finder ticket sales data was analysed in-house at the Audience Agency. The lead time of each booking (how long the tickets were purchased before an event) was cross-referenced by each of the following: sales channel (online or in person bookings), geographical region, Audience Spectrum segment, and art form by the twelve genres above. The sales channel analysis yielded little new insight, except to confirm that most tickets bought on the day of the event are bought in person. Following an interim comparison of findings, Audience Agency researchers ran two further analyses comparing lead time to average ticket yield and party size.
The two datasets offer radically different insights into lead time and the decision to attend in arts audiences; they each separately have their strengths and limitations. Audience Finder is unparalleled for its comprehensive capturing of ticket sales, in that it collects the universe of ticket bookings for partnering organisations, however, its reliance on ticket sales data means that non-ticketed events are under-represented, and many commercial, voluntary and small-scale arts events are not included in the system. The UACA interview dataset is unusually large for a qualitative study, providing in-depth exploration of audience members' subjective experiences of the arts, but its generalisability is still limited since the sample is formed of a small proportion of the arts-attending public. The datasets have different epistemological premises. Audience Finder is built on a positivist stance whereby audience engagement is investigated through observable behaviour (booking data), whereas the interpretivist approach of the UACA interviews centres on the subjective understanding of why and how participants make the decision to attend.
Despite their differences, there is value is analysing these two epistemologically different datasets together in this article. By holding both datasets in dialogue, each helps to unpack the findings of the other; Audience Finder data points to possible idiosyncrasies in the UACA interviews, which in turn suggest causes for the trends in the ticket sales data. As Mason (2006) has noted, the findings do not have to be "internally consensual and neatly consistent to have meaning" (p. 20); in the following section, findings from the two datasets are used to shed light on each of the four themes, at times supplementing, explaining, contradicting or saying something totally "outside" the other dataset. Audience Finder data shows largescale patterns in booking behaviour, within which the UACA interviews provide individual narratives, showing how the booking decision relates to broader motivations and lifestyle factors. In line with Mason's (2006) recommendations, we acknowledge the tension that exists between these two forms of data, and strive to produce "multi-nodal" accounts of our findings. This dialogical exploration of our two different ways of researching audiences offers new opportunities for quali-quant collaboration, analysing existing qualitative and quantitative datasets in such a way that each sheds light on the other and surfaces new insights into audience engagement.

Results and discussion
Audience Finder data showed that art form had the greatest impact on booking timelines, followed by: ticket price, the size of group attending, Audience Spectrum segment, and finally the geographical location of the ticket booker.

Art forms
In the Audience Finder database, the difference in lead times was pronounced by art form. Tickets for art forms such as Experimental Theatre and Physical Theatre sold much nearer to the event than events such as Rock & Pop/Hip hop (see Figure 1), with 87% of Rock & Pop/Hip hop tickets sold a fortnight before the event, compared to just 47% of Experimental Theatre tickets. Overall, theatre performances had shorter lead times than musical performances. It is possible that the key driver for longer lead times is likelihood of selling out, but this data is not collected in Audience Finder. The fact that Rock & Pop/Hip hop events had the longest lead time is unlikely to be a surprise to arts marketers or anyone who has tried and failed to acquire tickets to see famous music artists months in advance.
UACA interviews similarly showed different timescales of decision-making for different arts events. At one extreme, participants were aware that tickets for big-name artists (for example, Beyoncé and Kate Bush were mentioned by several participants) could sell out within hours, and it was necessary to be in the online queue as soon as it opened. At the other, by far the most spontaneous form of arts engagement were visits to art galleries, specifically those with free entry (paid-for exhibitions tended to involve more planning). The phrase "pop in" was used by 24 participants (18%) when describing gallery attendance combined with other activities such as shopping or meeting friends, and therefore gallery visits were discussed in very different terms to ticketed events. The flexibility of being able to drop in on a gallery at a convenient time was afforded by long opening hours and longrunning exhibitions. However, this meant that exhibitions were easy to miss without the commitment of a ticket, and some participants found themselves in a rush to see an Cinema trips were the second most spontaneous arts event, with participants commonly choosing to go to the cinema a matter of hours before the start of the film: "sometimes film's a bit like instant meals, it's just like: right, you want a bit of escapism, then what's on at the nearest cinema?" (Ld38). However, the rising cost of cinema tickets was forcing some participants to plan further ahead, and getting a ticket to a film on the door was dependent on it having a long run in the cinema; attending a one-off screening or watching a film on opening night could mean having to buy tickets in advance. For some participants, spontaneous cinema-going was seasonal, an activity reserved for dark winter evenings, whereas outdoor activities were more appealing in the summer.
The booking process for theatre was more complicated, largely due to the ticket price banding that most large theatres employ. Savvy customers were aware that the best deals were often available either as soon as tickets went on sale -"you can get much better seats for a much better price if you're organised" (Ld06)or by purchasing their tickets at short notice, through standby schemes, papering apps, or queuing for returns. For some participants, their experience was significantly affected by where they sat in the auditorium, leading them to book further in advance in order to secure their ideal seat. Since the UACA study was focussed on the contemporary arts scene, many participants attended theatre performances at smaller venues, seeing fringe or experimental works where they were more confident that cheap tickets would still be available "very last minute" (Ld34), in line with the Audience Finder trend for late booking for contemporary theatre events. For some participants, their preference for last-minute bookings therefore influenced the type of event they would attend.
Since higher ticket prices seemed to prompt earlier booking for cinema tickets and gallery visits above, we returned to the Audience Finder data to investigate the impact of cost on lead time (see Figure 2). We found a strong relationship between ticket yield and lead time: with each increase in price band, tickets were bought further in advance. With three months to go, 47% of £30+ tickets had already been sold, compared to just 9% of tickets under £5. Of bookings made on the day of the event, 52% of sales were less than £10 per ticket.
Cheap tickets may have facilitated more spontaneous attendance, since spending £30 on a ticket on the day of an event is not a decision made lightly for many people. However, it is worth noting that the longer lead times for higher priced tickets is the opposite of the pattern encouraged by dynamic pricing, where late bookers are charged more for equivalent seats. The connection between expensive tickets and long lead times may also be influenced by how much audience members want to attend an event, and therefore how willing they are to pay a premium for the ticket, and how eager they are to have the event in their calendar. High-profile events, such as those involving celebrities, may command a high ticket price and generate early bookings through fear that it will sell out. Therefore early and late booking can be seen to be influenced by audience members' perception of demand and availability of arts events.

Geography and availability of the arts
In the Audience Finder data, lead time varied considerably by region, with London and Scotland showing the shortest lead times, and audiences in the East and South East of England (outside London) booking furthest in advance (see Figure 3). This difference was most pronounced between a fortnight and two months before the event. The results from Scotland, Northern Ireland and Wales should be interpreted cautiously; whilst there are many bookings from these countries, most venues in Audience Finder are from England. This additional distance of travel can be expected to affect lead times for those places.
In the UACA interviews, the way in which London participants described their booking patterns differed substantially from participants in the other case study cities. Londoners were far more concerned about events selling out than in other cities (with the exception of rock and pop arena performances, as described above): "one of the problems in London is a lot of things you have to book miles in advance" (Ld19). London participants felt they had to book early because, while there were lots of cultural activities in the city, the large population meant that wellpromoted events were likely to sell out. One participant described how this issue of supply and demand could impact attendance in rural areas too: "occasionally Northern Opera would tour [to the Lake District and] it would be selling out in five minutes flat, because nothing came otherwise" (Ld39). The rarity of the event therefore influenced how far in advance participants felt they had to book.
At first glance, the UACA finding that Londoners felt they had to book in advance to secure a ticket appears to contradict the Audience Finder analysis above. While there were also more spontaneous London attenders in the UACA dataset, including people who used listings such as Time Out or Londonist in order to find events that were happening that same day, these examples of spontaneity were no more pronounced in London than in Liverpool and Bristol. Indeed, London participants in particular expressed a desire to be more spontaneous with their arts attendance, showing frustration at the difficulty of sifting through information to find something to do at short notice: Last Wednesday we found ourselves in the rare situation of going out for a drink after work and I was going, "let's go and see a band in a pub [a]round here". Two hours later, I was disgusted with the state of the Internet […] I know there's a dozen bands playing in a two-mile radius of here tonight, even on a Wednesday, but could I find that information?! (Ld27) Nevertheless, London participants differed in that they were more confident that there were many arts events taking place on any given night. The same participant remarked: "this is what's great about London, is there's stuff going on that I don't even know about yet". Participants from other cities were aware of the disparity of local arts provision in comparison to London, believing that they would attend far more arts events if they were living in the capital: "if I was in London and I had the money I would be out every night at the theatre. You are limited here [in Liverpool] to what is on" (Lv21). Believing there to be fewer arts events available in turn influenced what participants attended, with one participant in Bristol saying that "living in a city like Bristol, there's not a huge amount of places to go to, so I'll tend to go and see whatever it is they've got on, even if it's not necessarily something that totally piques my interest" (Br10). Participants' decisions to attend were therefore influenced by their confidence in the arts provision in their city, as well as the ease of finding out more information, and their perception of likely demand for tickets.

Socialising
Another factor that came out as a strong influence in booking patterns amongst UACA participants was the social context of attendance. Some participants had arts companions who booked tickets and organised attendance for them. Deciding to attend an event with a friend encouraged participants to commit to attendance, rather than making a decision nearer the time. Nevertheless, participants did also comment on how difficult it could be to get friends to commit: "I went to see my favourite artist in the world […] [and] there was like a 50% drop out rate amongst my friends" (Lv12). This was particularly difficult when an event was liable to sell out: "[I tell my friends] 'you need to tell me straight away because three days later the tickets will be gone'" (Ld34). Participants had developed strategies to manage this, such as: setting up online messenger groups to coordinate plans, only inviting friends to events where they could buy tickets on the door, or for giving friends a firm deadline to commit. Arts marketers have long been aware of the difficulties of organising tickets for a group, and have built in the option to reserve tickets and pay when numbers are finalised, however the use of online messenger groups to coordinate plans (and Facebook message's introduction of the "plan" function) suggests that younger audiences may be seeking out less formal means of organising attendance.
Previous research has often shown that "not having someone to go with" is a common barrier to attendance (Blume-Kohout, Leonard, & Novak-Leonard, 2015), and the difficulty of arranging to attend with companions prevented some participants from attending at all: I did want to go and see Othello […] [but] nobody is enthusiastic, and you think, 'Oh, I'll get round to it,' and then you suddenly find that there is only one date left […] [so] I just never got around to seeing it. (Lv36) Those participants who were happy to attend alone if no friends were interested, were still conscious of omitting the social part of the experience, sometimes through preference, and at other times because the desire to see the event outweighed the consideration of going alone: "it's never been a hindrance for me to go alone 'cause sometimes scheduling wise, it just doesn't work out" (Ld11). But however willing participants were to attend alone, they would not always "get round to it" because planning to attend with companions also made them commit to attendance, buying tickets or putting a date in the diary. Therefore, socialising at arts events seemed to increase attendance across the board.
Having found co-attendance to be a strong determining factor in the decision to attend timeline, we returned to the Audience Finder data again to analyse lead time in relation to the number of tickets bought (see Figure 4). Generally speaking, there was a correlation between the size of the group and lead time, with larger party sizes booking earlier. Attenders who bought one or two tickets were consistently the latest bookers, and increased group size correlated to earlier booking times. Audience Finder data therefore confirms the relationship between co-attendance and early booking found in the interview data. Nevertheless, co-attendance is only partially captured by quantitative data since audience members who book separately may still attend as a group, especially given the challenges of coordinating ticket purchases discussed above. This is a topic in which combining quantitative and qualitative data is particularly effective, shining light on the larger booking trends in the Audience Finder database, whilst also complicating that picture through the narratives of co-attendance in the UACA interviews.

Planners and last-minute bookers
Audience Finder data was analysed to see whether booking times varied for different groups in the Audience Spectrum segmentation (see Figure 5). Three segments bought considerably higher proportions of their tickets on the day: Metroculturals, Experience Seekers, and Kaleidoscope Creativity. One possible explanation for this is that these three segments are on average younger and more likely to live in London than the other segments, so this may point to late booking being a generational or metropolitan  phenomenon. Furthermore, Metroculturals can be characterised as cash-rich but timepoor, which may explain their tendency to book late since they may have limited flexibility in their free time. The greater variation by segment than region does, however, suggest that late booking is more influenced by profile than location.
Audience Spectrum is broken down into three levels of higher, medium and lower arts engagement. The three middle-engaged segments were found to book earlier, while the four lower-engaged segments booked later. Higher engaged segments varied: Commuterland Culturebuffs mimicked the booking patterns of the middle-engaged segments in booking early, whereas Metroculturals and Experience Seekers were far more like the four lower-engaged segments in booking late. This split amongst higher engaged segments again suggests that older segments book earlier and younger segments book later.
The UACA data also revealed a mixture of participants who liked to book early or late, although this was typically expressed in terms of preferences or perceived availability, rather than highlighting the geo-demographic basis evidenced by Audience Finder. Several participants noted that it was important to commit to attending an event, either through buying a ticket or scheduling it in their calendar: "you've gotta schedule [arts events] […] otherwise it just drifts into do[ing] a white wash and hang[ing] it up instead of going out and doing something" (Ld20). Other participants preferred not to book in advance, instead deciding to attend much closer to the event: "I live today and tomorrow and that's almost it" (Ld40).
There was a clear difference in how participants planned for events that were special occasions, in that they were eager to book early, to ensure that they had tickets and that the event was firmly in the calendar. This could be because there was a particular artwork that they wanted to see, because it was part of a trip, or because they were planning a social occasion: "we've already got our Christmas [theatre] with our children [booked] for next November … ! […] Otherwise you don't get everyone there" (Ld07). Participants spoke of wanting to strike a balance between planning and spontaneity, having some key events in their calendar, but going to other arts events more spontaneously. This ad hoc attendance could also take many forms, from stumbling across an arts event and deciding immediately to engage, to booking other events that caught their eye.
What came across particularly strongly was the idea of finding arts events as a way of meaningfully filling free time: "I went to the Philharmonic, it was a weekend and we didn't have any plans, so we just looked at what was on" (Lv27). When participants sought out events to fill specific periods of free time, the decision-making process was substantially altered. No longer did the programme dominate as the most important influence (Baker, 2000(Baker, /2007Brown, 2004); factors such as convenience, cost, ticket availability, and matching the arts event to their mood became important too: "do I want to sit back and relax, or do I want to see something off the wall?" (Ld05). These last-minute impulse buyers were less concerned about the specific arts experience than they were about filling an empty evening in their diaries. This finding contradicts the emphasis on cultivating audience loyalty in arts marketing, and suggests that there may be a segment of potential audience members who are willing to substitute a variety of different types of event if the time and price is right.
Socio-economic circumstances could determine whether or not participants were able to book early or late, which may therefore, be reflected in the geo-demographic differences evidenced in Audience Finder. Examples included: needing to save up to buy a ticket, having to organise childcare, having unpredictable health problems, needing to organise attendance around shift work, or having to plan to accommodate access requirements.
The only way that my family and I are going to be able to get tickets [for a blockbuster exhibition] is if we can find a day when we can go, and book them a week in advance; and that's hard, because you know, I'm on zero-hour contracts, I've got six jobs, […] I could be offered a shift, like, this evening; and at times, like, you have to take it. (Lv25) Access issues is the main thing, so […] I've always got to do my homework and there's always a compromise. So they'll say things such as, "Yes, we have a portable ramp that we could get out when you arrive". (Lv32) The UACA data reflected the distinction between intention and purchase in the Theory of Planned Behaviour (Ajzen, 1991) above, as participants spoke about putting events in their calendar or finding ways to keep longer-running events in mind until they could find an appropriate day to go. Facebook was described as a useful tool in keeping events on participants' radar, especially by clicking that they were "interested" in a Facebook event, which meant that they would be reminded about it until the day of the event. Booking late meant that participants could avoid over-stretching themselves when their "eyes [were] too big for [their] belly" (Br28). Delaying the commitment of booking tickets therefore allowed participants to decide whether to engage based on their mood, or whether other plans had materialised. Therefore, the late bookers identified in the Audience Finder data will consist of a mixture of audience members whose lastminute booking corresponded with a last-minute decision to attend, and those who had largely planned their attendance already, but had waited to purchase tickets.

Conclusions
This paper is a novel exploration into how new insights into the booking patterns of arts audiences may be revealed by analysing large quantitative and qualitative datasets in tandem. As stated above, the two datasets have been held in dialogue throughout, not presupposing any form of triangulation but acknowledging the tension between them. This dialogue has included corroboration (such as the finding that rock and pop events are booked earliest), contradiction (are London audiences early or late bookers?), and moments where the two datasets say something completely "outside" of each other (such as the geo-demographic trends in Audience Finder, expressed as personal preference in the UACA interviews). The Audience Finder data has shown the relative importance of factors such as art form conventions, ticket price, co-attendance, and geodemographic characteristics, and allows for identification of proportionately small differences with confidence through use of a large dataset. UACA data, meanwhile, has demonstrated the process by which some of these factors influence planning and spontaneity, placing the point of purchase in a longer decision to attend. Each of these types of analysis, therefore, illuminate a range of influences on the pre-attendance process.
The different approaches also complement each other's methodology, by providing depth and context, respectively, to enable a thorough analysis. One of the challenges of big data approaches is that the data collection occurs separately from and/or prior to the framing of the questions that it can answer. This type of combined analysis offers a means of formulating specific research questions that can be reapplied to the transactional dataset. Conversely, the volume of quantitative data allows for recognition of patterns which are not self-reported by interviewees. This can inform the framing and interpretation of the qualitative responses. While the UACA interviews were not designed to directly address questions from the Audience Finder dataset, they were nevertheless designed with the knowledge of this kind of robust sectoral reporting, and the qualitative design of the UACA project was intended to complement the strong quantitative research that takes place within the arts sector. Future research could extend this relationship further, by designing quali-quant studies that directly address gaps in knowledge in the Audience Finder dataset, or by testing the generalisability of key findings from the UACA study quantitatively on a larger sample of audience members.
The qualitative UACA data in particular suggests a three-part new model for understanding the decision to attend. Firstly, there are plans to attend arts events that are made well in advance, either through a desire to see a particular art work or performance, or in order to plan a special occasion. Secondly, there are moments where participants looked for an event to do on a particular day, which might be short term (this evening), or longer term (when a friend is visiting). In these situations, the actual programme diminishes in importance, as factors such as convenience, timing, or price come to the fore. Thirdly, there were participants who wanted to maintain certain level of engagement with the arts, and who would make decisions about what to attend based on availability and affordability, often talking about "getting the most for their money". This attitude may be more prevalent amongst UACA participants than the wider arts-attending public, since contemporary arts events were often cheaper and therefore provided more opportunities to attend for the same money. Since the UACA data demonstrated that individuals can, and frequently do, make decisions to attend based on all three parts of this model, this finding opens up new (albeit challenging) possibilities for targeted marketing to reach potential attenders at each of these three stages and so maximise the chance of interested audience members converting into bookers.
The three parts to this model demonstrate the importance of combining informationprocessing and hedonic models of decision-making to understand the decision to attend. The decision-making processes evident in the data often contain a blend of both rational and experiential motivations for attendance; for example, the prominence of spontaneous, last-minute attendance at art galleries is partly driven by the flexibility of both art forms, but this is also driven by the aesthetic and symbolic value of viewing art as opposed to getting a coffee, for example. This also relates to Hall et al.'s (2016) study of opera audiences, which suggests that a rational motivation, such as the additional benefits offered by a price bundling strategy, can increase consumer loyalty, provided the aesthetic and emotional values of the art works are not seen to be compromised.
Two findings in this analysis are particularly striking and deserve further discussion for their implications for arts managers. Firstly, last-minute impulse buyers were less concerned about the exact nature of their arts experience when looking for an event to fill their free time. As noted above, this seems a counterintuitive segment for arts marketers to target, since the emphasis in marketing is typically to cultivate loyalty amongst audiences. Nevertheless, these impulse buyers may be less particular about their arts choices than regular attenders, therefore may be more willing to take a risk for the right price and a convenient time. This points to arts events being consumed as an experience, chosen from a number of art and non-art options available to the attender, in which art takes on an equivalency to other recreational activities. In an increasingly experience-driven economy (Bronner & de Hoog, 2018;Pine & Gilmore, 2011) it may be worthwhile for marketers to conduct experiments in to how to maximise the appeal of their offering to last-minute attenders over other non-art options.
Secondly, London participants had considerably more confidence in the arts provision of their city than those in Bristol and Liverpool. The disparity in arts provision between London and the rest of the UK has been a long-standing debate (Stark, Gordon, & Powell, 2013), but this data suggests that the biggest challenge may be one of audience perception. Collective efforts to market the overall cultural offering of a city, through festivals, concerted efforts around City of Culture or similar schemes, or centralised online listings, may increase the self-confidence of audiences about their local arts provision. Cultivating the belief that there is "stuff going on that I don't even know about yet" (Ld27) in other cities may in turn increase engagement with the arts locally.
We have talked very little in this paper about the ways in which the arts can be inaccessible to certain segments of the population, however, our findings point to ways in which arts organisations can help to make their offering available to a wider population. Findings from the UACA interviews show the difficulty of engaging with the arts even for those who are heavily invested in arts experiences, and so while the views of those who reject, or feel rejected by the arts sector, are not represented in this sample, there are insights from our analysis which are relevant beyond the current arts-attending public. UACA participants displayed a high level of knowledge of how booking patterns work, such as knowing that an event is likely to sell out or that cheap tickets sell first. This knowledge comes from experience and needs to be made more explicit to less frequent attenders. Our analysis also highlighted how buying cheap tickets in advance was not always an option for some participants, whilst waiting to see if there were cheap tickets on the day was not practical for participants with children or audience members using wheelchairs, where it was necessary to plan ahead. Organisations could think more about when they release cheaper tickets, perhaps using dynamic pricing more flexibly, to make events more accessible to those Audience Spectrum segments who may not be able to book as soon as a season is announced. Further research is needed into whether dynamic pricing may therefore change the proportions of audience groups as much as the booking behaviour of individual attenders.
The combination of these two types of analysis make clear the importance of distinguishing between several phases in the pre-event process: awareness of an event, identification and selection of a particular event as one which may be attended, decision to purchase and making the purchase itself. However, the length of time spent in each of these stages is not uniform, with some attenders moving quickly from awareness to purchase, and others being unwilling or unable to commit until the day, despite intending to attend for a much longer period. This therefore raises the question: what tips someone over the edge to purchase a ticket and commit to attending an arts event? Our analysis suggests a number of factors: committing to attend as a group, making attendance a special occasion, and perception of scarcity and demand. There is a need to untangle audience members' desire to attend a specific art event from their desire to have an arts experience, regardless of its content. For some participants, the anticipation (or relief) or having arts events in their diary formed a part of their enjoyment; conversely for others, being spontaneous and not knowing what they might experience that week was an intrinsic part of how they lived their lives. These differences of decision-making approaches and timings, as well as the variety of factors that inform audience decisions summarised in the three-part model, show the complexity of the inter-relation between spontaneity and planning in arts attendance. Concurrent analysis of qualitative and quantitative data affords a richer view that either could provide alone, offering new insights into the ways in which the arts are valued and prioritised within audience members' lives. Notes 1. A full description of partner organisations, recruitment methods, research tools, participant statistics, and links to additional findings and publications can be found at http://www. sparc.dept.shef.ac.uk/uaca/. The 53 interviews conducted in Birmingham as part of the pilot phase but have not been included in this analysis, as they focussed on participants' lifelong history of arts engagement rather than how participants made the decision to attend. 2. Ethical approval for this research was granted by The University of Sheffield and included the assigning of participant codes to protect anonymity: these codes indicate the interview number and the city in which it took place (Br = Bristol; Bh = Birmingham; Lv = Liverpool; Ld = London).
video game and music industries, plus numerous commercial campaigns for global brands and charitable organisations. https://orcid.org/0000-0002-6714-7651 Oliver Mantell is Policy Research Director at The Audience Agency. Oliver delivers bespoke consultancy and research projects, using audience data to advise on marketing, policy and strategic planning. He has previously worked in a range of roles in arts consultancy, research, marketing and audience development (including work for Audiences Yorkshire/&Co and Museums Sheffield, and as a freelancer). He is a board member of Grimm & Co and Ignite Imaginations and has taught Arts Marketing and Creative Entrepreneurship modules at the University of Leeds.
James Trinder is a Data Analyst at The Audience Agency, spearheading new initiatives of delivering more complex analysis of the platform's "big data" and developing new data-driven tools for their clients, and the arts and culture sector. Over the last four years, James has worked on projects including: large-scale customer segmentation projects for Sadler's Wells, Opera North and The National Archives; feasibility studies and market appraisals; and research for London Borough of Culture, UK City of Culture, and British Association of Concert Halls.