Fighting ISIS on Facebook – Breaking the ISIS Brand Counter-Narratives Project

By Anne Speckhard, Ardian Shajkovci & Lorand Bodo

This blog post synopsises a study that attempted to intervene with over fifty English-speaking Facebook accounts endorsing, promoting, and following ISIS.

Methodology and Research Design

For this study, several anonymized Facebook accounts were used to identify English-speaking radicalized Facebook profiles as our target for an online counter-narrative intervention. All the Facebook users admitted into the sample were operating in the English language, and most appeared, based on their postings, to be UK residents.

This study was exploratory in nature. A multi-level research approach, each phase developed in succession, was used to explore the digital environment on Facebook. In the process, we attempted to identify the most suitable method to counter ISIS’ ideology on Facebook and to learn if our counter-narratives could reach the target audience and if so, make a positive impact that we could measure, or at least observe.

In under a week, we were able to assemble a sample of nearly fifty English-speaking accounts that openly endorsed, followed, and shared ISIS propaganda. Posts, comments, shares, and likes served as metrics to identify those we considered as extremist and ISIS supporters online.

The second step in the process involved the intervention itself, which was planned to be two video uploads of the ISIS defector ICSVE Breaking the ISIS Brand Counter-Narrative video clips The Glorious Cubs of the Caliphate and A Sex Slave as a Gift for you from Abu Bakr al- Baghdadi. Additionally, the sample was tagged in the video to promote the video being watched.

The final step involved analyzing the results of our intervention. At this final stage, we were handicapped in our efforts as our accounts were disabled by having them unfriended, and possibly reported to Facebook by those with whom we were intervening .

In the second stage of the intervention video clips were focused tested with highly committed and radicalized individuals, who were selected on the basis of already being members of a closed ISIS-following and endorsing Facebook group. One of our original four anonymized accounts had managed to penetrate this group—interestingly by invitation after befriending radicalized account holders. By friending our fourth account, that one was also allowed in. Our other two anonymized accounts had been disabled at that point due to our first foray.

This second approach inside a closed-ISIS supporting Facebook group allowed us to test the results and potential impact of posting the ISIS defector video clips in the ‘echo-chamber’ of account holders who were already quite serious in their endorsement of ISIS. We expected them to be unlikely to be shaken by any counter-narrative materials given their commitment to ISIS, but we tried it nonetheless to observe their responses.

The last approach was designed to target so-called “fence-sitters” We defined them as those who not only “liked” ISIS-related Facebook pages and content, but also “liked” pages that also represent and/or disseminate liberal and democratic viewpoints, etc. Our assumption for this sample was that these individuals had not entirely narrowed their focus to ISIS and similar extremist groups and might be more amenable to an intervention.

In terms of research ethics, the research was conducted with a careful consideration of ethical concerns related to data collection, legal concerns, confidentiality, protection of research participants’ identities and data storage. Our research was also rooted in ethical guidelines for academic research, including following strict guidelines on Internet research with vulnerable extremist individuals online.


The data analysis yielded two important conclusions. A majority among the target audience seemed to have watched the beginning of the video clip, but not the rest of it. Seven Facebook users whose profiles were marked as highly radicalized (they showed open and very supportive attitudes towards ISIS) liked the videos. Given the nature of the video (i.e. openly depicting the harsh realities of living under ISIS rule), one could assume that these seven users had not actually watched the entire video but instead were responding to its ISIS thumbnail picture and ambiguous title, which could be seen as pro-ISIS.

For instance one of these seven, a female user commented to another user: “And if the police come knocking, where do I say I was with you?” This statement underlines the assumption that the sample did not watch the complete video clip, but rather assumed by its title that it was ISIS content and therefore liked it. This supports the argument that giving the counter narrative clips pro-ISIS titles is a good strategy for reaching the target audience. The user who commented on the video also unfriended the anonymized account when she discovered her mistake in endorsing it.

Similar to the results in the first approach, it seems that the videos were not watched in their entirety by highly radicalized and committed individuals in the closed ISIS chatroom, although we are not certain on that score. This becomes especially apparent in the closed news feed group that was infiltrated by the two anonymized accounts. To post the video, one of the eight admins had to approve the video. Basically, that policy ensured that no one can post content that violates the group’s “policy” or, in this case, their extremist beliefs. Nevertheless, the defector video was successfully uploaded into the group and, most importantly, it was approved once again by one of the admins, reaffirming the importance of using pro-ISIS names and possibly also ISIS thumbnail pictures on counter-narratives to reach the target audience. It was interesting to see how easily the admin approved content —or how the videos managed to slip through—that directly contradicted what the group was initially set up.

The third approach with fence-sitters yielded different results. They were more likely to watch the whole video compared to the other two approaches. This became apparent through comments made and emotions expressed (e.g. sad or angry faces). Comments were made in this group criticizing the authenticity of the videos—the first ever in global focus-testing that questioned the authenticity of the defectors appearing in the video. Arguably, this implies that the video had been watched, including the parts in which the defector denounces ISIS, as in comparison to the first two approaches in which account holders simply liked and shared the video and when watched entirely, questions about authenticity may have been the only way to attack the videos as the content is too damning if the speaker is authentic.

Gender issue were an interesting aspect of this online intervention study. We quickly learned that the use of male and female accounts was necessary, as some female accounts only accepted requests from “sisters,” with the same holding true for males only accepting other males as Facebook friends. It is assumed this reflects adherence to conservative Islamic principles of modesty and propriety. We respected these notices and opted not to test whether these account holders actually did as they said.

Posting as male users yielded fewer responses from those receiving the post (who were both male and females as our posts were shared with a much wider group than simply those we tagged) compared to posting as female accounts. By using male accounts, videos were simply liked and shared—without even watching the video (approach one and two). On the other hand, when posting from a female account, comments such as “You have nothing to do with Islam […] If you think hijab is not fardh [Islamic obligation] then you become five stars kafirah [unbeliever]” were made.

In this respect, it is important to mention that the profile picture used depicted a young girl from the back who was not wearing a hijab, which made an opening for the possibility of attack. It could be argued that the comment was directly targeting the issue of not complying with religious beliefs. This also implies that the use of a male account to post is much simpler in terms of posting counter-narratives as it overcomes issues of appropriate female dress in the profile depiction. However, using a female account yielded more reactions in terms of likes and views compared to using a male account. It is also important to mention that we received approximately forty message requests to open and start a conversation. We believe these messages were mostly flirtatious in nature. We declined to get into conversations in this study.

Nevertheless, using female accounts was more effective in terms of disseminating counter-narratives than male accounts measured in terms of likes and views. We did not test if women talking to women, or women talking to men, made a difference but intend to do so in future studies.

The purpose of this exploratory study was threefold: to learn if we could reach ISIS supporters and endorsers on Facebook (our target vulnerable audience) and learn about their reactions and responses to ICSVE-produced counternarratives. The results of our online intervention indicate that we have managed to infect the ISIS-dominated sphere on Facebook. We also learned that our counter-narrative videos, at least partially, were like, shared, and watched and obviously created some level of engagement causing viewers to think about the messages contained in them.

Discussion and Conclusions

Our primary goal was to reach individuals who are further into the radicalization trajectory. If our Facebook target audience were serious ISIS supporters and promoters, any engagement with our video material on their part can be considered a measure of success, mainly because at the point they are on the terrorist trajectory they will have likely narrowed their focus down only to material coming from ISIS, or groups like ISIS.

While we were able to establish a certain measure of success as far as our intended objectives, we lacked sufficient data to ascertain exactly how our counter narrative videos were received by our target audience, as we were unable to monitor the long-term effects. We also managed to entice ISIS endorsers into opening, and even endorsing, and sharing our counter-narratives due to the videos pro-ISIS names and appearance. That said, we could not observe or report any changes in users’ cognitive or behavioral changes. It may be that those who are already deep into ISIS ideology, particularly those who are in prison or in ISIS territory, are unlikely to be swayed by these videos (however in our offline testing we have found those categories of individuals to be influenced by them). Those just dabbling in ISIS, however, may be able to be turned back. Given we found that many of the friends of those we intervened with also clicked on our videos this is an important consideration—that we reached not only our target audience but influenced their circles as well. We do know from other research carried out by ICSVE, that the counter-narrative videos deeply affected an ISIS emir in prison who watched them, so we cannot rule out that highly committed ISIS cadres would not be affected by their reaching them via the Internet. Clearly, more research is warranted in this respect.

Similarly oriented research with radicalized individuals online indicates the possibility for reversions from forward motion on the terrorist trajectory by using Internet- based interventions. Our study followed in the footsteps of such studies, and in many ways replicated their findings, namely confirming that such individual could be reached and engaged.

The online effects of our materials are still not well understood. Difficulties especially arose in the form of our inability to continually observe the results of our postings. We continue to produce and test ISIS defector counter-narratives both in offline and online settings. We firmly believe that studies such as these represent a unique opportunity to reach out to individuals who might not otherwise seek help, or who might seek help in all the wrong places. We also remain hopeful that our interventions online could potentially lead to more creative ways of working to reach and redirect those already on the extremist trajectory towards better alternatives.

Equally important, while our Breaking the ISIS Brand Counter Narrative vidoes represent an important step in discrediting both ISIS and its ideology, including redirecting vulnerable populations online away from the threat of terrorist groups like ISIS, reducing terrorism and appeal to terrorist groups like ISIS will also require addressing underlying psycho-social, economic, and political issues that serve as both push and pull factors for such vulnerable populations to consider terrorist groups like ISIS as a solution to such contentious issues in the first place.

Anne Speckhard, Ph.D., is an adjunct associate professor of psychiatry at Georgetown University School of Medicine and Director of the International Center for the Study of Violent Extremism. Follow @AnneSpeckhard on Twitter.

Ardian Shajkovci, Ph.D., is the Director of Research and a Senior Research Fellow at the International Center for the Study of Violent Extremism.

Lorand Bodo, is a former Research Fellow at the International Center for the Study of Violent Extremism. Follow @LorandBodo on Twitter.

Leave a Reply