Social media – definition and history

Social media is over two decades old (Figure 1), having begun with the website Sixdegrees.com. The current most popular tool, with 2,006 million active users,5 is Facebook.

The originator of the term ‘social media’ is contested,6 and sometimes contradictory terminology has been applied. Obar and Wildman7 define social media as:

• using (currently) Web 2.0 applications
• dependant on user-generated content (USG) for continuing existence
• a service which creates (or facilitates creation of) a user profile
• a service which facilitates the connections of an individual to a community online.

Previous research

Library researchers began to study social media in the mid 2000s. A search (on 10 May 2019) of Library and Information Science Abstracts for peer-reviewed articles on ‘Social media’ AND (Library OR Libraries) netted 4,124 results. Short of a full systematic review,8 any study is restricted to a snapshot of the scholarship. In addition, social media is in a constant state of flux, and articles become outdated fast, for example by mentioning defunct tools such as the six-second video tool Vine.9 This desire for currency means that the scholarship seems to be more article-, blog- and web-based.

Researchers10 have parsed the literature around social media use by libraries into chronological phases, beginning with the initial discussion and distaste for social media, progressing into enthusiasm and uptake, then analysis of tools and usage, and finally into issues with the new medium. However, a research area barely two decades old does not need to be parsed chronologically. Instead, it is possible to classify the research by content into the following genres:

• advice for libraries using social media (‘how to’)
• measuring the success of social media (‘appraisal’)
• case studies using social media (‘what happens’)
• dialogues about social media and some limited criticism (‘why bother?’).

Inevitably, a few researchers produce work which fits into more than one category,11 but it is still a useful classification.

‘How to’ literature

Representing the largest genre of literature, this section focuses on tutoring the ‘at work’ librarian, providing a methodology for implementation,12 sometimes refined for specific types of library13 and focusing on elements such as marketing14 or specific tools like Twitter.15 Others16 eschew discussion of specific tools for a more holistic approach to using internet-based information gathering.

‘How to’ literature formalizes best practice for social media use. A good library social media account may be considered to fulfil the following five criteria:

1. is active17
2. utilizes the tool fully and correctly18 – for Twitter this would involve using hashtags, handles and other formalized, searchable terms in a user’s bio19 and tweeting using images20
3. considers its audience21 – knowing who, what and where their audience is
4. has a strategy for their social media22 including, perhaps, an official policy23
5. uses tact24 and style – being considerate of their audience and the open nature of social media.

However, it is acknowledged that the standard model may not fit well for smaller teams,25 and we may ask whether libraries who have experienced ‘success’ also follow these criteria.

‘Appraisal’ literature

This type of literature addresses ROI for libraries,26 outlining the recording and analysis of metrics27 and key performance indicators (KPIs)28 to determine success or failure. While warning of the dangers of using social media as a ‘quick fix’ for marketing,29 researchers espouse its virtues for smaller libraries.30 Some discuss more nebulous ideas for measurements such as the idea of a ‘relationship currency’31 mentioned in the introduction.

‘What happens’ literature

Containing reflective case studies,32 surveys33 and statistical analysis of social media usage, this type of literature provides a snapshot of specific areas of the social media landscape. Analysis is limited, e.g. to one or two specific libraries,34 or restricted by tool.35 Overall, there is a distinct bias towards academic libraries.36 This literature provides conclusions, questions and methods which will be discussed and utilized in this study.

Despite the large number of ‘best practice’ works, some of these case studies have concluded37 that a ‘one size fits all’ methodology does not work and that libraries would be better served formulating their own strategies. Some conclusions reached imply that use of social media requires little training and indeed, that this is one of the advantages.38 Survey respondents have also concluded that social media is perceived to be a low-cost option for libraries but gathering metrics to analyze ROI is difficult.39

There are few studies that compare different types of library40 and analysis of intended audience and participation often nets contradictory findings.41 Many are restricted to a single type of library42 and conclude that social media does not in fact save costs as it can often take a substantial time investment43 across multiple tools44 for true success, contradicting the time-saving benefits espoused in appraisal literature.

Three case studies are particularly notable for having collected and analyzed large quantities of data. Crawford’s45 study of US public libraries criticizes the Twitter follower metric46 and warns of the dangers of empty profiles,47 noting that larger libraries often reach a smaller percentage of their users than smaller libraries.48 He is one of the few researchers to discuss in depth auditing49 and the closing of accounts upon bad ROI.50 Mon and Lee’s51 study into Twitter builds upon Crawford’s52 earlier work utilizing ‘new key metrics emerging for assessing social libraries on Twitter in terms of visibility and self-representation, reciprocity, audience, activity, and influence’.53 This paper discusses how earlier users of Twitter became ‘power users’54 and concludes that there is a ‘strong correlation’ between visibility on Twitter lists curated by Twitter users and libraries’ PeerIndex (now known as Brandwatch) scores.

Al-daihani et al.’s55 analysis of Twitter concludes that ‘regularly updating and maintaining the library Twitter feeds (attracts) more users’56 and there is no relation between the number of tweets and the number of followers. This study will attempt to corroborate this conclusion.

Shulman57 measured not followers, but the influence of a Twitter network, finding that institutional networks ‘such as those maintained by the college/university, departments, and programs’58 have the highest influence on Twitter, despite not having the highest number of followers or the most tweets. Shulman’s study is of interest for the originality of its methods, if not for the scale of libraries studied (only two). Its interesting conclusion that influence, not followers, should be the true measure of success was discussed with the smaller libraries in this study.

‘Why bother’ literature

More discursive in nature,59 this small genre discusses legal issues,60 ethics61 and the pros/cons of social media use62 while debunking myths.63 Again, there is an academic bias in this section.64 There was one example found of analysis and criticism of social media,65 which asserts that people, especially academic library staff, are not ‘asking the right questions’. Instead, libraries are overwhelmed by positive portrayals of social media and are not viewing it objectively.

Summary and studies used

The literature on social media cannot be summarized neatly and has not existed long enough for chronological trends to be apparent. Instead, it can be divided into thematic groups. The largest, instructional ‘how to’ literature formulates universal principles for best practice. Other literature outlines case studies, surveys and statistical analysis of the current state of social media use, often restricted to a tool or type of library. Few papers outline the appraisal of ROI for social media and provide discursive pieces about the changing face of libraries, ethics and law with regards to social media.

The review of the literature reveals several key areas for further analysis. A lack of cross-discipline library studies has given the literature an academic library bias. Social media has been deemed a time-saving device for smaller library teams or, contradictorily, a significant investment across different tools. The smaller libraries contributing to this study were questioned on their uptake of ROI appraisal methods and their use of multiple tools for social media. Finally, there is limited literature available on the legality, ethical and objective analysis of social media use in libraries that is not overwhelmingly positive, and this study therefore attempted to be more objective.

Stage 1: creating a list of UK library Twitter accounts

A list of UK library Twitter accounts had to be created as one did not exist at the start of this project. The lists used here may not be comprehensive, and there is currently no comprehensive list of libraries in the UK. Considering that not all libraries will have a single account, or have a social media presence at all, names and metrics were gathered from a range of sources and used to estimate numbers.71

Five categories were created for this project: Academic/Higher Education (Acd/HE), School/Further Education (Sch/FE), Medical (Med) and Public (Pub). Public libraries were further segmented into accounts posting for an entire county (Pub/CC), a London borough (Pub/Borough) and any combined or joint accounts (Pub/Joint), for example @leedslibraries and @HarrisPreston. A final category, other, was created to encompass miscellaneous libraries including arts/heritage and law libraries. Using the available metrics, a required number of accounts for each type of library was assigned: 500+ for Pub, 100+ for Med, 200+ for Acd/HE and 200+ for Sch/FE.

A Twitter account (@SarahHLib) was created, with Twitter lists for each type of library. A general search was made for ‘UK Libraries’ and the term ‘Library’ with a geographical restriction to the UK. Some of the lists and metrics were used to formulate other search terms, such as the names of Oxford libraries or London boroughs. Pre-existing Twitter lists72 and the ‘who to follow’ function on Twitter also revealed several overlooked accounts.

The gathering was impeded by several factors. Twitter’s internal search has little nuance, making it difficult to create meaningful search results, and many libraries do not include the word ‘library’ or ‘libraries’ in their account name or description to make discoverability easier. Furthermore, some of the names used on Twitter can be misleading; school libraries may use the acronym ‘LRC’ (Learning Resource Centre), whilst NHS medical libraries may use their trust acronym. Location data is not an audited field and similar names caused confusion – for example, the Canadian ‘London Library’.73 Every attempt was made to counteract these impediments. (See Table 7 for number of libraries.)

These lists were exported into Excel using Twlets,74 which extracted the account name, location data, time/date created and basic Twitter metrics, such as number of followers. As the information gathered was (and is) publicly available on Twitter, it was deemed unnecessary to anonymize the data. Data was extracted from Twitter on two convenient occasions: 12 June and 24 July 2017. As with any social media study, the data quickly grows stale, and Dave Rowe, working with the Libraries Hacked group, has created a site75 should anyone wish to replicate or use a current version of this data, though random checks have indicated that results will be similar to those in this study.

Stage 2: defining smaller libraries

This article is specifically interested in Twitter use by smaller libraries. The definition of a ‘smaller library’ will vary depending on the type of library. For example, a university library with a large collection but few staff will have similar challenges to a smaller school library with a single member of staff. Crawford’s76 study ranks the size of library dependent upon the size of the catchment area for users. However, this method cannot be applied to libraries with no numerically defined catchment area, such as special libraries or international libraries. In a more specific, type-restricted study, stricter criteria could be used.

For the purposes of this study, a smaller library was defined as one which:

• represents a single institution (nt national libraries or an account for many libraries posting with one ‘voice’)
• has staff numbers that are below the average for that type of library77
• has a low staff-to-collection size ratio (discovered through direct enquiry to the library).

Stage 3: interviews

Five libraries were selected (one representing each type of library) using a list ranked by follower count (Table 1), with the top ‘smaller’ library contacted for interview. Interviews were used as a more insightful method into the libraries’ actions concerning social media, as the interviewer could steer answers in a manner they could not in questionnaires.

Two SurveyMonkey ‘forms’78 were sent to these libraries along with a request for permission for the information to be used in this study. The first form asked for details about the library and their social media presence including other tools, such as Facebook, for comparative purposes. The second form guided the library through the taking and recording of basic metrics on their Twitter accounts, using the following services:

• Twitteraudit79 – takes a sample of 5,000 followers and analyzes their accounts, calculating if said accounts are bots or other ‘fake’ accounts
• Twocation80 – reveals where followers are located by percentages, which are indicative of declared non-audited locations
• Twitter analytics – retrieves metrics regarding tweets, tweet impressions, profile visits, mentions, top mention engagement figure and most popular tweets and engagements.

Each participating library was sent a list of questions, and interviews were conducted over the internet using Appear.in81 and recorded using Audacity.82 After editing, a transcription was sent to the library for them to verify that the quotations and information were correct.

Twitter use by different types of library

Libraries registering on Twitter

The study located 1,423 unique library Twitter accounts in the UK. The first three had registered accounts in 2007, with a further ten having registered in 2008. The surge of registrations began in 2009 and has remained steady since (Figure 2).

Figure 2 

Number of libraries registering on Twitter by year

Types of library with Twitter accounts

Different types of library are not evenly represented on Twitter (Figure 3). Despite the literature focusing on academic libraries, this study found public libraries (when taken in their totality) to be the most common.

Figure 3 

Number of types of library with Twitter accounts

Key to types of library:

Pub/Joint: public with any combined or joint Twitter accounts; Pub/CC: public posting for an entire county; Pub/Borough: public posting for a London borough; Pub (Misc.) – other public; Sch/FE: School/Further Education; Acd/HE: Academic/Higher Education; Med: Medical

Different types of library took to Twitter at different times (Figure 4). The first libraries to begin registering en masse were Acd/HE libraries, peaking in 2009. Public libraries have also been registering since the beginning, although registration did not peak until 2013. Med and Sch/FE libraries did not begin registering until 2009. Thus far, Med libraries appear to have reached their peak in 2015, whereas Sch/FE peaked in 2014, but registrations have stayed high. The ‘other’ category remains low, peaking in 2010 during which year 12 ‘other’ libraries registered.

Figure 4 

Library accounts registered per year by type of library

Key to types of library:

Acd/HE: Academic/Higher Education; Med: Medical; Pub: Public; Sch/FE: School/Further Education

Are libraries on Twitter following best practice criteria?

Two of the five best practice criteria identified earlier – whether an account is active and whether it uses the tool fully – can be assessed using this data set.

Criterion 1: is active

Of the libraries on Twitter, six have never tweeted at all. A further 134 have not tweeted for six months and were deemed inactive. Approximately 90% of accounts can be considered active, having posted a tweet in the last six months (Table 2).

Table 2

Number of libraries by last year tweeted (up to 2017)

Last tweet year	No. of libraries
Never	6
2009	2
2010	8
2011	8
2012	12
2013	11
2014	12
2015	23
2016	53
2017	1,288

The average number of tweets per day by UK libraries was shown to be 1.48. Pub/CC libraries have the highest average number of tweets per day, posting approximately 3.56, followed by the other public library accounts for joint libraries (2.64). The lowest rate of tweets is from Sch/FE libraries, who post 0.92 tweets on average per day (Figure 5).

Figure 5 

Average number of tweets per day by type of library

Key to types of library:

Sch/FE: School/Further Education; Pub: Public; Med: Medical; Acd/HE: Academic/Higher Education; Pub/Borough: public posting for a London borough; Pub/Joint: public with any combined or joint Twitter accounts; Pub/CC: public posting for an entire county

What are the key influencers on follower count?

The most rudimentary test for success on Twitter is the number of ‘followers’, with ‘likes’ representing account interaction. The data gathered here can be used to verify suggested influences on follower count.

The average number of followers per account (Table 7) was 1,381, falling by 12 in the six weeks between data gatherings. However, follower numbers varied considerably, from 1 to 51,501. ‘Likes’ per account average out to 612 per account.

The average number of followers was found to be quite variable between libraries of different types. The highest category was ‘other’, with 5,460 followers on average, followed by Pub/CC with 2,763, Pub/Joint with 2,551, Pub/Borough with 1,984, Acd/HE with 2,318, Pub with 901, Med with 582 and Sch/FE with 287. Most of these averages saw an increase in total followers between the two data gatherings, although Acd/HE and Pub/CC showed a decrease of 23 and 102 respectively. This affected the overall public library average, causing it to decrease by five. The number of likes per account also varied wildly depending on the type of library, though the highest and lowest average were also ‘other’ (with 1,327) and Sch/FE (with 429).

Activity

The activity of an account also has a dramatic effect on follower numbers. Out of the 1,423 libraries, 1,288 (approx. 90%) posted at least once in 2017. These accounts had a higher follower count than accounts that had not posted so recently (Figure 6). However, even accounts with no tweets whatsoever (six accounts) still had some followers, averaging around 109.

Figure 6 

Average number of followers by most recent tweet

Key to types of library:

Acd/HE: Academic/Higher Education; Med: Medical; Pub: Public; Sch/FE: School/Further Education

Time since account registration

A further major factor influencing follower count, which has not previously been recognized or analyzed, is the amount of time an account has existed. The average number of followers gained per year was calculated (followers divided by year fraction) for every account in the data set. Averages for each type of library were then calculated.

Discounting 2007 (when only three accounts registered), the average number of followers gained per year falls from 4,730 for accounts registering in 2008 to 98 for accounts in 2017, a decrease of 4,632 over nine years (Figure 7). There is a distinct downward trend regardless of library category (Figure 8).

Figure 7 

Average follower count over time

Figure 8 

Average followers by type of library over time

Key to types of library:

Acd/HE: Academic/Higher Education; Med: Medical; Pub: Public; Sch/FE: School/Further Education

Case study interviews: how smaller UK libraries measure and achieve ‘success’ and ROI when using Twitter

To gain a more detailed understanding of the use of Twitter by smaller UK libraries, this study interviewed five institutions with high followings for their type and size. The libraries were asked to capture metrics for discussion and were questioned regarding criteria two, three and four in the best practice criteria. (See data declaration for location of transcripts and case studies.) Criterion five – uses tact and style – was not assessed in this study due to a lack of data to enable a sentiment analysis like Sultan’s83 to be made.

Types of library using Twitter

Libraries are now well established on Twitter, with the number of registrations per year peaking in 2009. Some libraries which should, theoretically, be on Twitter (Waltham Forest, Barking and Dagenham Borough Libraries and Bournemouth University Library) are not. Interestingly, early academic library uptake of Twitter appears to have significantly shaped the literature itself and the approach to Twitter.

Analysis of best practice

Despite an impressive amount of best practice advice available, many libraries are still not utilizing the potential of Twitter to its fullest extent. Some accounts are not using Twitter’s in-built formatting tools to improve the discoverability of their ‘Twitter bio’, lacking descriptions, location data or even searchable keywords such as ‘library’. Accounts are mismanaged and sometimes not used at all – despite warnings from researchers86 regarding misusing social media. There were, however, also examples of good practice evident, representing libraries utilizing the tool to its fullest extent. Examples of all-encompassing best practice are in the minority, so it cannot be said with confidence that good practice = successful accounts, especially if using the simple metric of Twitter followings.

Effect on followings

The variation of followings according to the type of library is significant. School and medical libraries have smaller followings than academic and public libraries, though there are underachieving accounts in all types, with the variation being most pronounced in public and academic categories. Therefore, the type of library appears to influence growth in followers, and this should be borne in mind when setting social media goals and measuring success. Further research could be made into how the content of each type of account influences this.

After the type of library, the core influence on follower counts is time. Followers grow over time, regardless of activity. While an active account will have more followers than an inactive one, even abandoned accounts can have hundreds of followers.

Crawford’s87 criticism of the follower metric and Al-daiani’s88 conclusions regarding number of tweets and followers is corroborated by the data in this study, though the daily number of tweets is less important than ensuring the account is active. This research has shown that the number of followers is dependent on the date on which an account was created and on the type of library, rather than on best practice. Newer accounts may have ‘missed the boat’ for high follower numbers, skewing the perception of success when using the blunt comparison of follower counts.

Are smaller UK libraries using best practice?

The five libraries contacted for this study have large followings and a sound grasp of how to use Twitter, though they could be using more tools, etc. to exploit it fully. They are cognizant of a strategy but remain willing to alter things, allowing for a ‘guidelines not policies’ approach to social media structuring.

On paper, social media seems to be a win-win situation for smaller UK libraries, representing a new, faster, cheaper platform for more traditional library work in marketing, outreach and publicity. The five libraries interviewed demonstrate a more complicated reality as their audiences go beyond their obvious user groups. This presents a problem currently under addressed in the body of research on library social media use; if the libraries are not connecting with their user group, who are they connecting with? Does it even matter? Or is it, like St Bride asserted, about the evidence of interested people, not the nature of the interaction?

Social media is not just for basic outreach and publicity;89 it is for collaboration work, interaction with global users, current awareness (be it with parents, scholars or staff), resource gathering and for telling people who will never visit or interact with a library that it exists. Social media is a defence against invisibility for smaller libraries, allowing libraries previously sidelined by their geographical location or collection type to highlight what they do and allow people to assist them in their endeavours, cashing in their ‘relationship currency’.90

How do smaller libraries measure success/ROI?

The libraries studied have devoted a significant amount of time to their social media accounts, with some such as Orkney spending time out of work hours to perfect their Twitter. This refutes some91 conclusions while supporting others,92 testing the boundaries of the concept that social media is ‘cheap’.93 Once you discount the initial ‘followers = successful’ equation, success becomes more nuanced. Some libraries attribute success to the number of retweets and likes, others to authors agreeing to visit the library. Yet, despite knowing many examples of ‘success,’ none of the libraries have recorded these experiences. This creates some concerns for the future of these accounts, as they lack the KPIs94 that would allow them to note a fall in ROI.

Study critique and further research

The main limiter of this study was the choice to restrict social media observation to Twitter. Twitter was chosen due to the accessibility of information and the large number of libraries using it. Finding the same metrics for Facebook would be more complicated due to the private nature of certain groups. Analytical tools exist for Twitter which have the capacity to gather large amounts of publicly available data, making it an ideal tool for a dedicated study of UK libraries. As a tool, Twitter presented some challenges with regards to ‘real’ followings, true location data and the libraries themselves being difficult to find due to duplicate accounts, mismanaged accounts and inactive accounts.

This study was originally conceived to analyze several social media tools. However, the workload involved proved too great, especially as there was no consolidated list of UK libraries. Further research into creating these data sets would be useful. Active data for the 1,500+ libraries now on Twitter is currently available95 and further analysis could perhaps investigate a sentiment analysis of types of tweets or the influence of an account using Shulman’s96 methodology or Mon and Lee’s97 PeerIndex scores.