Verification Agencies on TikTok: The Case of MediaWise and Politifact

Abstract

This research aims to analyse the work of two international information verification agencies on TikTok—MediaWise and Politifact—according to their evolution, approach, content, and format. To this end, a quantitative approach has been used with an inductive content analysis with nominal variables, which offers specific nuances adapted to the unit of analysis. In a first phase, an empirical analysis was carried out, focusing on the measurement and quantification of the number of publications and interactions of the audience, from the time Fthey started operating on this platform until 31 December 2023. The total number of posts extracted was N > 704, which generated N > 4,166,387 user responses. In a second phase, an in-depth content analysis of all the posts published by these two agencies in four months (October and November 2021 and October and November 2023) was carried out, allowing us to analyse their evolution, but also to compare the two agencies in terms of approach, themes, and style. The most important findings show that both agencies adapt the style and narratives to this social network through the use of dynamic resources, a casual and informal tone, and elements of humour. In addition, both contribute to public reason through different strategies: MediaWise focuses on media literacy and Politifact on verification, using resources, effects and content in line with that purpose. Finally, we observe a downward evolution in terms of reach and impact on the audience, as well as a lower dynamism in 2023 than in 2021, which opens the door to future lines of explanatory research that delve deeper into possible causes.

1. Introduction

TikTok is currently the most successful social network for the exchange of audiovisual content if we take into account the following four variables: highest number of downloads on mobile phones globally for three consecutive years [1]; acquisition of a large number of new users in the last four years [2]; advertising revenues in 2023, which exceeded the sum of those of META (Facebook and Instagram), X (Twitter), and Snapchat [3]; and the increase in annual time that (mainly generation Z and Alpha) spend on the platform [4].

Likewise, in the analysis of its rapid success, it is possible to cite various scenarios that have worked in its favour as well as business actions undertaken by the Asian company that created its success. Thus, it can be admitted that the pandemic period during COVID-19 boosted TikTok’s growth as well as that of other companies in the entertainment industry; especially for TikTok, the pandemic occurred at a time of global expansion [5,6,7]. Another highlight is the sophistication of its algorithm, which is one of the most advanced and complex operating on the Internet [8,9]. Moreover, it is the only social network to date that places the sophistication of its algorithm at the pinnacle of the audience experience it generates [10,11,12]. Equally, the algorithm also achieves another effect: its users stay online longer, consuming more content [4,7]. Similarly, one could argue that their business strategies are a reason for this feat. TikTok initially focused on short (15-s), dynamic videos to appeal to a younger audience. However, they have gradually opened the door to longer and longer content, from 60 s to 180 s (July 2021), going up to 10 min in 2022 [13,14]. While TikTok has argued that with this opportunity, they expect an increase in user creativity [15], they have also threatened YouTube’s ground with this initiative [16], as new and more diverse users are arriving on a monthly basis in response to the call for increasingly heterogeneous content. While Generation Z—those born between 1997 and 2012 [17]—are still the largest age group on this social network, since the pandemic, generations, Y and Z have started to become more visible [18]. Consequently, these strategies of broadening the diversity of its audience and allowing the possibility of hosting videos with more varied themes and durations are bearing fruit [19,20]. Finally, it should be noted that video, as an audiovisual format, is clearly relevant in a wide range of contexts [21], and, especially at present, it is arousing great interest among young people [22] as the increase in video production and consumption has been astronomical in recent years [23]. This has also helped TikTok to grow as a social network for sharing audiovisual content.

The new generations prefer the audiovisual format, even for consuming news on social networks. According to the Reuters Digital News Report 2023, the increase in news consumption on TikTok is beginning to be significant [24]. Similarly, the Pew Research Center has published a study of US adults that confirms that one in three in the 18–29 age group follow the news on TikTok. This is not only true for young adults; in addition, 15% of adults aged 30–49 follow the news on TikTok. This figure rose by 5% from 2022 to 2023 [25], being a significant proportion of this age group. Additionally, this type of news is what is known as snackable [26], as it is short-form content that is viewed in a short amount of time.

2. An Overview of Published Articles

There is an abundance of work that explores the most common strategies for impacting audiences on social networks. Park et al. [27] analysed the importance of news article headlines and descriptions on Twitter for being shared by users, such that clickbait-style tweets with a link to an article increase user engagement. Additionally, news items with a high emotional component are shared more than more neutral ones [28]. Other authors [29] conclude that negative sentiment increases the virality of tweets.

Some proven strategies for impacting young people are diversifying the news offering, experimenting with new forms of storytelling, fostering journalists’ communication skills, and increasing interaction [30]. In the specific context of TikTok, Salb [31] focuses his research on the use of text and other elements on this social network. The inclusion of text (closed captions or reinforcing messages) presents numerous benefits that are enhanced in the case of users with a native language other than English. This author [31] also argues that the first video or image that appears may influence the likelihood that viewers will watch the entire piece.

As the traffic of informative videos has grown on TikTok over the last few years, different initiatives have emerged to verify the information shared on this platform, so that fact-checking agencies have become key players in this context of ‘infoxication’ [32] as well as one of the ways to combat growing misinformation [33]. In this sense, fact-checking has also been adopted by the media to counteract the effects of the spread of false information [34]. The content of fact-checking agencies is related to current affairs and news, so the findings of Boczkowski et al. [35], according to which younger populations consume news on social networks incidentally rather than through proactive and targeted search, are applicable.

We can define fact-checking as “the verification of data using current tools, including new technologies” [36]. In the context of growing distrust in the media and the proliferation of false or misleading information, fact-checking becomes a valuable accountability mechanism and an opportunity for journalism to regain its social relevance [37].

Snopes.com, created in 1995, is considered the first fact-checking initiative [38], although it was not until the new century that the first verification agencies focusing on politicians’ statements appeared. FactCheck.org was created in 2003 and Politifact (Poynter) in 2004, along with a verification programme that the Washington Post also launched in that period. All of these initiatives were responses to the presidential elections of the same year [39]. The creation of the International Fact-Checking Network (IFCN) by the Poynter Institute in 2015 was a point of no return, as verification agencies would begin to proliferate from this point onwards [40]. Furthermore, it is important to note that the best-known models of verification agencies either rely on conventional media or follow an NGO model [41].

The most important trends in fact-checking over the last decade have been a shift towards a global movement, the importance of the COVID-19 pandemic and the 2016 US elections in the growth of online verification, the growing importance of digital content verification versus politicians’ statements, and the increasing prominence of visual versus textual content [42]. This last point is vital, given that, according to this author [42], in 2016 agencies only verified 5% of social media content; now, in 2021, this figure has risen to 66%. In this regard, Lucas Graves has recently made an important distinction between fact-checking and debunking [36].

The objects of fact-checking are statements by politicians, often in mainstream media, whereby the raison d‘etre would be to promote public reason: an objective and complex debate in which nuances are explained and the topic is explored in depth [36]. Debunking, however, deals with extreme content shared on social networks by anonymous subjects, so the ultimate goal it pursues is public health: the aim is to disprove false information quickly in order to stop ‘contagion’ and virality due to the risks they pose to society, but without the need to explain or elaborate on the subject [36]. In their own words [36] (p. 14):

“A tension between informing audiences, and protecting or even managing them, emerges in the strategic concerns around amplification and online backlash which attend debunking work, superseding traditional news values to a degree. Debunkers balance the traditional imperative to inform against a heightened sense of responsibility to minimize the potential harms of information, based on the understanding that “fact-checking can do harm if you do not choose well what to fact-check”.

Related to this, it is important to address the importance of pedagogy and media literacy in fact-checking [30,31,32,33,34,35,36,37,38,39,40,41,42,43]. Some authors, such as Lee and Ramazan [44], consider media literacy one of the ways in which we can improve the health and information maturity of citizens through fact-checking. A close link is therefore drawn between the work of fact-checking agencies, media literacy, and the proper functioning of democracy [45]. In fact, many fact-checking agencies have sections (or even specific focus) on media literacy, education, collaboration with young or vulnerable audiences or, in general, a pedagogical approach in their verification work and subsequent communication to the public. Verification agencies use social media as a means to disseminate their findings. In their research, Dafonte-Gómez et al. [46] conclude that all agencies use Twitter and that Facebook also plays a prominent role, with YouTube being the third most used platform. However, their study does not include TikTok, although there are 42 agencies in Europe and America with active accounts on this network [47]. In this regard, McCashin and Murphy [48] point out that there is a mismatch between the scarce literature and the growing importance of TikTok, highlighting the relevance of conducting quantitative and qualitative content analyses to work with TikTok content [48].

Some works have analysed the use of TikTok by verification agencies. For example, Lu and Shen (2023) analyse the activity of Chinese verification agencies on TikTok, including three areas of analysis: audiovisual, persuasive strategies, and engagement metrics. The results show that fact-checkers’ posts tend to have higher brightness, less use of cool colours and a faster pace than other accounts’ videos, as well as the use of five persuasive strategies: humour, logic, storytelling, authoritative sources, and clickbait thumbnails [49].

The work by López et al. [50] compares the use of TikTok by agencies in the US, Brazil, Spain and Portugal, with the aim of identifying the different disinformation practices in each country. Sidorenko-Bautista et al. [51] also compare themes and general information from different international agencies, including surveys of agency users. Along the same lines, Arrieta-Castillo and Rubio Jordán [52] delve deeper into the narratives used, the topics covered, and the impact achieved by the posts of Ibero-American verification agencies.

On the other hand, there is previous research that has addressed one of the two selected agencies, Politifact and MediaWise. In the case of MediaWise, Çömlekçi [53] highlights the importance of literacy education, considering it an ‘unorthodox’ agency. Other authors, such as García-Ortega [54], reinforce this pedagogical characteristic of MediaWise.

Some works stand out, such as that of Lim [55], who compares the results of the Politifact and FactChecker verifications and concludes that they do not coincide in the verified claims (only 10%); he concludes that there is coincidence in the results when there is an absolute diagnosis (true, false), but not when they are ambiguous (misleading; true, but...; false, but…). These conclusions are confirmed by the research of Markowitz et al. [56], as the results confirm that the findings of different agencies agree when they are absolute (true, false), but not in between. In line with this research, Diep [57] compares six independent (including Politifact) and media fact-checkers during election periods.

Much of the scientific literature about fact-checking agencies discusses their characteristics, their verification methods, their most common practices, and their distinguishing features or provides an overview and literature review of fact-checking agencies [37,58]; the audience for hoaxes on social media [59], the use of labels to classify fact-checks [60]; the impact on the audience (Jang et al., 2019); or the challenges of debunking [61]. However, there is little literature on the activity of these agencies on the social network TikTok.

The justification for this research, therefore, stems from the need to develop an academic corpus commensurate with the social, media, and informational prominence that both TikTok [50] and fact-checking agencies have achieved. Secondly, we know that young people are one of the audiences most vulnerable to disinformation [62,63] and that TikTok, as we have shown, has become a reference platform for this audience when it comes to receiving information.

Moreover, disinformation and the virality of false information not only pose a risk to the users themselves who use the social network but also pose a direct threat to the proper functioning of the media and democratic system [64]. In this context, the work of verification agencies on social networks, especially on TikTok, becomes particularly important, both in terms of media literacy and verification as a tool to stop the viral ‘contagion’ of false and harmful content.

3. Materials and Methods

3.1. Research Methodology

This research analyses and compares two reference verification agencies in terms of evolution, approach, content, and format. Secondary objectives derive from this main objective: (O1) to analyse the growth, reach, and impact of these two accounts on TikTok; (O2) to identify the purpose of their publications and their main themes to confirm that there is coherence with the agencies’ priority objectives; and (O3) to find out the resources they use to design their videos to see if they are in line with the TikTok narrative.

The specific research objectives of the research are specified in three research questions, which facilitate the identification of the what? of the research [65] in addition to presenting the problem in a straightforward and less distorted way [66]. The research questions answered by this research are as follows:

-

RQ1: How have the main TikTok accounts of verification agencies in the United States evolved in terms of impact and reach?

-

RQ2: What are the main thematic and formal characteristics of these two accounts?

-

RQ3: What kinds of strategies and resources do they use to adapt to TikTok, its young audience, and the work they do?

3.2. Sample

In order to achieve these objectives and answer the research questions, descriptive research was carried out using a quantitative methodological approach that combines different techniques appropriate for both the social network and verification agencies. The selected sample was obtained from two international information verification agencies. This selection was based on four criteria: verification agencies in the United States that are part of the International Fact-Checkers Network, that frequently use TikTok as a verification tool, and that have a community of at least 100,000 followers. The two agencies that meet these requirements are PolitiFact (@politifact) and MediaWise (@mediawise), with more than 150,000 followers, recurrent activity on TikTok, and accounts created more than three years ago (Politifact) and four years ago (Mediawise). The unit of analysis will be the publications (videos) of these accounts on TikTok.

3.3. Data Analysis

In the first phase, an empirical analysis was carried out following the consolidated work of previous research on media and communication on social networks [67,68,69,70]. This first analysis focused on measuring and quantifying the number of publications and audience interactions with their content since they started operating on this platform until 31 December 2023. The total number of their posts extracted was N > 704 and generated N > 4,166,387 responses from their audience. These interactions have been divided into three metrics that have also been studied separately (likes, comments and shares) as they themselves offer other types of information that enriched the study. Finally, some of the formal elements that make up the messages were studied to find out whether the platform’s resources are being optimised. Thus, the use of hashtags was studied, as well as the number of characters used in the caption accompanying the video. The raw data for the TikTok accounts used in this research were provided in Excel format by Analisa.io, one of TikTok’s official data providers. Unfortunately, this company has unfortunately ceased to operate in the market since January 2024.

In a second phase, a content analysis was conducted, following the definition of Mayntz et al. [71], because it is considered the most suitable option for addressing the content of the social network TikTok [50] and has been widely used to analyse TikTok [6] and to address the use of TikTok by verification agencies [50,52]. Furthermore, this applied methodological approach was designed and tested through previous research carried out for this purpose: Content analysis methodology for fact-checkers on TikTok [72].

To ensure the consistency of the sample and the variables of analysis, inductive coding was followed as proposed by Salbs [29] and Lookingbill [73], according to which an initial coding of 20% of the sample and subsequent grouping into a series of variables, most of which were nominal, were carried out; this allowed specific nuances adapted to each unit of analysis to be obtained. Specifically, the final coding scheme resulted in 24 variables: fact-checking agency; verification; media literacy; section; topic; presenter; gender; age; humour; music; subtitles; pop-up labels; video; pictures; screen recording; screen browsing; screenshot; underlining and circles; zoom; text effects; emojis; graphic design; animation; and final claim.

Given that our objective was to compare both agencies and analyse their evolution, the sample chosen comprised the total number of videos published in two specific time periods. Specifically, all posts published by Politifact and MediaWise over four months (October and November 2021 and October and November 2023) were analysed. The selected analysis period included the most recent posts available in 2023; selecting the same period in 2021 made it easier to analyse the evolution, and dealing with the same time period for both agencies facilitates comparison. The content analysis comprised an in-depth analysis of the entire sample (N = 76), which allowed us to attend to the specific nuances of each agency and, therefore provided more detailed information on the phenomenon studied. Furthermore, content analysis is considered an ideal methodological approach for communication research [74].

4. Results

4.1. Results of the Account Analysis

Although Politifact started operating on TikTok just one year later (August 2020) than Mediawise (August 2019), Politifact’s success on this platform—if one pays attention to the variable of followers—came earlier.

In November 2022, this research began collecting data on these accounts on the TikTok platform. The variable “followers” is a metric that can only be obtained by constantly monitoring the account month by month, as no marketing company can find out what the number of followers was in a particular month in the past on TikTok. It can only be known in the present, at the time the data are requested. Thus, in this research, it was possible to find out what the data were in November 2022 because the study was started at that time.

Table 1. Followers of TikTok accounts.

Followers	MediaWise	Politifact
November 2022
	24.900	155.900
November 2023
	155.200	166.400
December 2023
	155.600	166.300
January 2024
	156.900	166.400

Source: Author’s own material.

shows the development of the acquisition of followers, and it is striking that although Mediawise started earlier on TikTok, its number of followers in November 2022 was very low (24,900 followers). This is not the case for Politifact. On this date, they already had 155,900 followers. This figure was mainly reached between the end of 2021 and the end of 2022, as their activity, as will be seen below in the number of publications, became regular during the summer of 2021. However, in just one year (from November 2022 to November 2023), Mediawise grew to 155,200 followers, i.e., an increase of 523.29%. This rate has been decreasing in the most recent period, with a growth rate of 1.10% from November 2023 to January 2024. The case of Politifact is different. It showed a rapid rate of follower acquisition during 2020 and 2021, but in 2023, it seems to have slowed down. Thus, this account has had difficulty in acquiring a new audience in recent months.

Publications on TikTok and their conversion into audience views are two of the variables that were also analysed. In the case of MediaWise, the coverage of the US presidential election in November 2020 that it provided can be seen in Figure 1. As a result, the average number of monthly publications rose sharply during September and October of the same year. Beyond this, its annual publications have doubled since 2021. Its average number of weekly posts rose from 1.3 to 3 posts per week in 2023. However, despite the increase in annual posts, views decreased between 2022 and 2023 by 78%. This is revealing and will be discussed in more detail below. Politifact also increased its annual publications. It went from publishing an average of 1.1 messages per week in 2021 to 2 messages per week in 2023. Thus, Politifact presents the same case as MediaWise, as its content views on TikTok have decreased by 82% compared to 2022. Moreover, while both companies suffer from the same issues—their audiences are dropping despite an increase in weekly posts—Politifact’s reach, i.e., the number of views per message posted, has been higher than MediaWise’s since 2021. Specifically, its reach was 68% higher in 2022 and 82% higher in 2023. These data support the idea of a possible higher engagement of your audience with your content.

The three audience metrics analysed were likes, comments and shares. Since 2021, in which Politifact showed regularity in publishing content on TikTok month on month, the data it obtained in all variables were superior to MediaWise.

Table 2. Type and number of interactions per year.

Likes	2019	2020	2021	2022	2023
MediaWise	194	228.296	42.334	436.010	116.449
Politifact	0	4.586	123.506	2.924.985	186.598
Comments	2019	2020	2021	2022	2023
MediaWise	10	4.381	1.530	9.966	1.951
Politifact	0	642	3.773	34.045	9.767
Shares	2019	2020	2021	2022	2023
MediaWise	14	6.711	799	4.215	925
Politifact	0	244	2.335	18.387	3.731

Source: Author’s own material.

lists these data, which also show, beyond the greater loyalty of Politifact’s audience to its publications, some remarkable details. Firstly, in 2022, Politifact achieved a record number of interactions. This is mainly due to the verification work carried out on four political content videos that went viral in January, February, April, and July of the same year. Two of them exceeded 700,000 likes, and one of them generated more than 8000 comments and was shared more than 9000 times. As a result, the interactions achieved by Politifact are difficult to surpass or repeat.

Another detail that can be extracted from these data (Table 2) is the drop in audience interactions in both agencies during 2023 and that these data are aligned with the visualisations of the previously analysed published content. Finally, it is worth discussing the accompanying text that these companies insert on the TikTok platform when they upload videos. This space is technically called the “caption”. It includes some introductory words along with mentions of other accounts or hashtags that categorise these publications. The characters included in these captions have been computed since they started operating on TikTok, and an evolution has been seen. Following the same line of thought, 2021 is taken as a year in which both accounts operated normally on TikTok. Thus, an increase in the number of text attachments accompanying posts was detected. Especially remarkable is the growth from 2022 to 2023. Politifact showed an increase of 103% and MediaWise 107%. This fact shows the importance they have started to give to more complete and detailed captions.

Likewise, another component of these texts is hashtags. Although this type of tag was invented by Chris Messina in 2007 on Twitter [75], these indexing techniques are now global, are applied in very diverse networks, and enable a large number of uses [76,77,78], the most common being the possibility of creating lists where all the content catalogued with that idea or concept is accumulated. Additionally, hashtags have been found to provide users with more effective communication, in part because of the implicit denotation in the tag that avoids ambiguities [79] and therefore also facilitates recall. They are a tool widely used for marketing and advertising purposes to sell brands and build loyalty.

Table 3. Most frequent hashtags published by Politifact and MediaWise during 2022 and 2023.

Politifact 2022		Politifact 2023		MediaWise 2022		MediaWise	2023
#learnontiktok	80	learnontiktok	190	#factcheck	97	medialiteracy	192
#LearnOnTikTok	33	factcheck	100	#medialiteracy	74	factcheck	177
#factcheck	21	fyp	74	#learnontiktok	44	learnontiktok	77
#COVID19	18	Biden	28	#tiktokforgood	23	tiktokpartner	41
#biden	10	COVID19	21	#tiktokpartner	23	tiktokforgood	40
#politics	9	Trump	16	#misinformation	20	election	31
#vaccine	9	politics	13	#election	20	misinformation	28
#roevwade	7	debate	12	#Trump	19	climatechange	24
#facts	7	Florida	12	#facts	12	checkyourfacts	24
#foryou	7	vaccine	12	#2020election	12	trump	22

Source: Author’s own material.

shows the tags most repeated by both accounts since they started on TikTok. It has been previously mentioned that in 2022, this same information was obtained for Politifact and MediaWise, and therefore, these data will make it easier to find an annual growth figure but also to determine possible qualitative changes in their use. Firstly, the increase in their use over twelve months is remarkable. If you take the most repeated label in 2022 with a match in 2023, you can find the percentage growth in twelve months. To this end, it can be seen that #learnontiktok on Politifact increased by 137.5% and #factcheck on MediaWise by 82.4%. These figures show an interest in reaching more individuals by indexing hashtags. In addition, it can also be seen that MediaWise makes greater use of hashtags, and this has possibly been one of the keys to its success in acquiring 523.29% more followers between 2022 and 2023. Likewise, the incorporation of a hashtag that did not appear in 2022 and that Politifact does not use was also detected in this account: #medialiteracy. At the beginning of the 21st century, the concept of media literacy began to be defined academically, being understood as the result of the process of teaching and learning media, i.e., the knowledge and skills acquired by students through critical thinking (Buckingham, 2005). Nowadays, media literacy is becoming increasingly relevant in the classroom and is therefore constantly a subject of academic debate. From 2003 to 2023, the number of indexed publications dealing with this topic increased a 100-fold [80]. MediaWise’s new strategy goes in this direction; its messages are linked to the idea of encouraging critical thinking by educating its followers. Politifact also displays hashtags that hint at the account’s commitment to learning and education if you look at #learnontiktok or #factcheck, which are also used by Mediawise; however, this account has a wider variety of hashtags related to political or media issues.

4.2. Results of the Content Analysis of the Publications

There are important differences between the two agencies in the purpose of their publications, and this is related to their strategy and positioning on the platform. Politifact prioritises the verification of false information without an explicit intention to focus on media literacy. Thus, 92.1% of the videos analysed confirm this result. In contrast, media literacy within MediaWise is much more visible in its work. In 2021, 93.3% of videos were related to some form of media literacy, and in 2023, 87% were related to media literacy. Their most common media literacy technique tends to be a post in which a final media literacy-related tip is included.

The posts of the two agencies can be classified according to the format used (presenter, pictures with text, dramatisation, etc.). In both cases, the inclusion of a presenter is the most frequent option in both agencies, with a predominance of young women, especially within MediaWise. Likewise, Politifact‘s strategy has a similar structure: a presenter in front of the camera who addresses the user directly and in which a live verification is carried out. A representative example of this format in both agencies can be seen in Figure 2. This script was repeated in 80% of 2021 posts and 88.8% of the 2023 posts analysed. The remaining 11.1% of the 2023 posts consisted of a succession of static images with written text in which different verifications are compiled. In the case of the 2021 posts, the remaining 20% consisted of humorous posts.

The same is true for MediaWise. Some 70% of posts in 2023 and 80% in 2021 had a similar script: the narrative driver looks into the camera and explains the verification of the content, during which time media literacy tips and recommendations are offered, or both. However, 21.7% of the 2023 publications had a different structure. These pieces were generally shorter (between 9 and 12 s) and showed a written text, without a presenter but with media literacy tips.

A comparison of the topics covered by the two fact-checking agencies showed that in 2021, still within the pandemic, health topics predominated in both agencies (61.5% of MediaWise and 56.6% of Politifact). However, in the thematic evolution experienced by MediaWise in the two periods chosen for this analysis, publications on social issues abound, where we found videos related to Halloween, missing children, rumours, celebrities, etc. Similarly, it can be seen how health issues began to lose prominence in favour of others related to the environment, society, politics, or Israel’s war in Gaza. The theme of health in Politifact follows the same dynamic analysed in MediaWise. However, political topics (13.5%) and the conflict in Gaza (33%) rose considerably. Thus, it is clear from the analysis of the themes that Politifact has a more political focus, while MediaWise has a more social focus.

In this section, it has been noted that both agencies use different resources to make it easier for the user to understand and follow the video, such as subtitles, pop-up labels, or the inclusion of videos and/or images to reinforce a narrative. In that sense, there are notable differences between them, but also between the two periods analysed. Politifact has shown an evolution since 2021, as subtitles were included in almost all posts (88.9%) and supporting videos were used in 77.8% of the posts analysed in 2023, compared to 5% in 2021, as can be seen in Figure 3. These videos are used to support narration, to reinforce certain messages, or as a resource that gives dynamism to the post.

In the case of MediaWise, pop-up labels were used to emphasise the text, but subtitles were rarely used, as shown in Figure 4. MediaWise also relied on the use of images and videos to support their narrative, choices that followed an upward pattern in 2023. However, MediaWise did not show major differences between the resources used in 2021 and 2023.

The inductive coding did not include the clickbait category because no significant use of this strategy was found. Likewise, storytelling and logical reasoning resources appear in practically all the videos of the two agencies and follow the most repeated format (a presenter verifying or explaining a piece of information). The use of effects in the videos predominates in both cases, as this option offers greater dynamism to the pieces and is aligned with the narrative of the rest of the videos on TikTok, where it is very common to find them [70].

As for the effects used by MediaWise, it is worth highlighting the use of techniques that favour the pedagogical work they carry out in their posts (Figure 5). In this way, they use screen recording, screen tracking and screenshots, as well as reinforcing the narration with circles, underlining, zooms and text effects that appear on the screen. These resources achieve coherence between form and content, as they make it easier for the viewer to follow the verification process and assimilate the key steps to detecting false or misleading information (carrying out a search, investigating the author of the information, or resorting to official pages, etc.). Thus, it can be seen that in both periods, the use of screen recording (73.3%) and, especially in 2023, the use of screenshots increased (30.4%). In addition, in 2023, graphic-type resources followed an increasing pattern, including emojis (from 20% to 30.4%); animation (from 13.3% to 17.4%); and graphic design (from 13.3% to 17.4%).

Politifact, however, does not use many of the above-mentioned didactic resources to verify its content. For example, screenshots were used in only 50% of the posts in 2021 and screen recording in 50% of the posts in 2023 (Figure 6). On the other hand, the use of animations and graphic design resources stands out.

However, it is worth noting that Politifact in 2023 opted for an animated formula that showed the result of the verification and is shown in Figure 7, a resource composed of animation and graphic design. In line with this analysis, it can be seen that their decisions on the use of these techniques pursued their main objective, which is to verify information, rather than a pedagogical purpose.

The register in MediaWise’s use of language is characterised by informality. It showed closeness to the receiver, both in the 2021 and 2023 posts. In the first period, 87.5% of the videos used elements of humour. They generated this comedy through different resources, such as the presenter’s comical appearance in the scene, puns, and the use of music or sounds. Although 2023 continued with the same light-hearted and informal approach, the use of humour in its pieces decreased drastically to 12% of the videos analysed. In the case of Politifact, the register is also informal, but somewhat more serious than that of MediaWise. Politifact also has elements of humour in its posts, although in a residual and decreasing way: 15% of the videos in 2021 resorted to humour, compared to the period analysed in 2023, in which no explicit element of humour was included.

Finally, the ending of videos from each agency follows a coherent line in accordance with its main objectives. Mediawise almost always closes with a media literacy tip. However, as noted above, the number of videos containing this dynamic decreased from 2021 (75%) to 2023 (58%). Similarly, it is worth noting that 87% of Politifact publications in 2021 included a summary of the verification. This has been replaced by a closing that appeals to audience interaction on TikTok. This new dynamic appeared in 89% of videos in 2023, with the following appeal: “see a claim we should check, tag us in the comments”. In this sense, Politifact has also evolved towards more interaction, as 78% of the posts analysed in 2023 were responses to users, compared to 5% of the posts analysed in 2021.

5. Conclusions

This research aimed to analyse and compare two reputable verification agencies on TikTok to understand the work they are doing and their impact on the audience that follows them. The first specific objective was to analyse the reach and impact of MediaWise (a project created in 2018) and Politifact (a project created in 2007) on this social network for sharing audiovisual content. This first aim was related to research question RQ1, which asked about the growth and evolution of these accounts on TikTok. Although MediaWise started operating on TikTok a year earlier, their efforts and resources were enough to reach a consolidated community of followers from the beginning. The crucial period for this agency was 2022, during which they increased their audience by 523.29%. This substantial increase could have obtained due to the incorporation of more didactic explanations in their videos, as will be explained below. The case of Politifact is different. They started their work on TikTok a year later; however, they were able to consolidate a large follower base within a few months of arriving on TikTok. This scenario was mainly due to: the greater power of the Politifact brand, which has been operating since 2007 and had already consolidated its name, unlike MediaWise; however, in addition, the importance of their political theme has to be stated. This theme engages Gen Z and millennials and mobilises them quickly [81]. However, although they have an important community, they have found it difficult to grow in this most recent period. This is not the case with MediaWise, which, although its growth has slowed down, has followed a continuous upward pattern. Similarly, although both increased their average number of monthly posts since the start of TikTok, they are not achieving stability in audience responses, which are very irregular. This suggests that one of the main elements impacting the success of their posts with their followers is the choice of topics to check. If these have proven to have achieved a certain virality, the audience’s responses will follow the same pattern.

In reference to RQ2, which asks about thematic and formal characteristics, it can be confirmed that both fact-checking agencies have built their own identity on TikTok and have been evolving with the optimisation of the resources of this social network without forgetting the specific roadmap that characterises each one of them. Similarly, it can be confirmed that MediaWise is adapting more closely to TikTok’s own narrative in terms of both content and format. This adaptation is attracting mostly Generation Z users. In this sense, Politifact is also evolving towards a search for greater dynamism through the use of graphic resources and humour. In this context, it should not be forgotten that the main purpose of Politifact, as this analysis has shown, is to verify false or misleading information, while MediaWise focuses on media literacy, as other authors have stated in their studies on this agency [53,54]. Moreover, the fact that both depend on the same organisation, the Poynter Institute, reinforces the specific focus of each of them. From this perspective, it is possible to understand and make sense of the characteristics analysed in the content analysis. Therefore, both respond to the NGO model, which is closely linked to the Poynter Institute.

Finally, RQ3 investigated the strategies and resources used by both agencies to address one audience or another. Thus, it can be confirmed that MediaWise, with its choices, seeks a younger audience and adapts the style and content of its posts to the more dynamic character of this social network and its characteristic audience. In this way, it can be observed that social issues are the ones that appear most frequently, and their presenters tend to be younger. Moreover, the use of formal resources is closely linked to facilitating the content, as many resources consist of directing the audience’s gaze to the screen and transporting them to a master class with a teacher. A more dynamic style is observed in this agency as it uses very diverse resources within the TikTok narrative, together with the use of pop-ups instead of subtitles. There is also a greater use of humour and in general a more casual tone. On the other hand, Politifact’s target audience is broader and its tone, in general terms, is more serious. Following this line of thought, the political theme has a lot of sway over this agency’s posts. Likewise, the use of resources and video effects is more moderate than in the case of MediaWise and aims to verify information and help the user to synthesise the most important information. Therefore, the use of subtitles and graphics that summarise the result or the conclusion at the end of the post are elements that make it easier to understand and remember the verification carried out. In this sense, the contributions of Lu and Shen [49] regarding persuasion strategies are supported, as the use of storytelling and logical reasoning was observed (as well as, to a lesser extent, the use of humour and an informal and casual tone). No significant use of clickbait strategies was found. However, there was extensive use of text as a reinforcement element, the effectiveness and benefits of which have been explained by Salb [29]. Politifact, in 2023, included subtitles in virtually all its videos, facilitating understanding for both native and non-native speakers [29].

In terms of evolution between 2021 and 2023, both agencies maintained a characteristic style that did not change much over the period analysed. However, some differences can be observed. Politifact showed a greater evolution in that it incorporated or changed some dynamics. Particularly noteworthy is the inclusion of subtitles in most posts in 2023 (88.9%), the use of videos to reinforce a narrative (77.8% in 2023), and animations and graphics to reinforce the result (77.8% in 2023). In the case of MediaWise, the evolution was more moderate, maintaining broadly the same type of effects and style. Some graphic resources (emojis, animation, etc.) increased in 2023, as well as the use of effects on the recorded screen. On the other hand, there was a slight decrease in the number of posts with media literacy content and fewer posts with comical or humorous elements.

The joint analysis of all the variables analysed showed a high degree of coherence between the themes, the style, the resources used, the most frequent hashtags, and the focus and purpose of the two accounts analysed: media literacy, in the case of MediaWise, and online verification, in the case of Politifact—both canonical and fundamental functions of verification agencies. In relation to the most recent and relevant research on fact-checking [36], there has been a clear effort to contribute to public reason (a calm, fact-based social and political debate), although Politifact also carries out important debunking work to curb the virality of online hoaxes, which is considered a public health approach [36]. The two agencies have different but complementary approaches: MediaWise focuses on educating citizens, especially young people, by providing them with tools to verify news; Politifact focuses its efforts on verifying false information, often with political content and aimed at a broad audience. In both cases, the agency contributes to improving the quality of public debate.

In that sense, the results are unambiguous and consistent with previous research. In future research, it would be relevant to complement this descriptive research with explanatory research that explores the causes of the decrease in the impact and interaction of both accounts, as well as the reasons why both accounts use less dynamic and humorous resources in 2023 than in 2021. In addition, this work will enable new lines of research comparing the activity of other international agencies on TikTok that operate in different regions.