Future Audiences/Generated Video

Future Audiences is exploring strategies to expand beyond our existing audiences of readers and contributors so we can truly reach everyone in the world as the “essential infrastructure of the ecosystem of free knowledge.” This includes investigating ways to reach young audiences and provide them the opportunity to learn from Wikimedia projects, as well as a path to become Wikimedia editors.

Building on last year’s research and insights on short video platforms and creators, who often get their information from Wikipedia without telling their readers this is the case, we are beginning an experiment with generating short videos – starting by repurposing existing community content (English Wikipedia “Did You Know?”s), and measuring how these videos perform on TikTok and other short video platforms (Instagram Reels, YouTube Shorts).

Experiment FAQs

Why short video?

We see experimenting with publishing short videos on popular short video platforms as an opportunity to understand how to connect with a large global audience of young people who are not already visiting Wikimedia projects.

• According to our yearly Brand Health surveys, young global internet users report lower awareness and usage of Wikipedia than older individuals, meaning there is a share of young people who aren’t getting information from Wikimedia projects.
• Among young people, short video platforms continue to grow as a popular destination not just for entertainment, but for news and educational content.
• Last year’s social video creator’s research indicated that there are many popular and successful short video creators who a) create edu-tainment content on a variety of topics, and b) often get their information from Wikipedia without telling their audiences that this is the case.

Is there a risk of damaging Wikipedia’s brand or enriching a non-free company’s brand by publishing content on TikTok?

TikTok is a very popular platform among young people, which is our main motivation for experimenting with publishing content there. Like all large for-profit online platforms today, it also faces serious criticism and regulatory pressure for unethical practices and harms. It is currently banned in India and is under threat of being banned in the United States. However, this has not deterred a large number of well-known media companies, nonprofit organizations, politicians, and other notable individuals and organizations globally from establishing a presence on TikTok. We believe that the risk of not taking the opportunity to learn how young people might engage with our content in new ways – as these other organizations have done – outweighs the risk of reputational harm from publishing content there. We are also not solely focusing on TikTok and are also planning to experiment on other social media platforms, including Instagram Reels and YouTube Shorts.

What did we learn?

How can we reformat Wikipedia content into short videos? What tools would be most useful in scalable content creation?

Our team created a tool that leverages AI, Wikipedia’s knowledge library, and media from Wikimedia Commons to create short-form videos. We also enlisted the help of both internal and external experts to create guidelines for our content’s format and subject matter. Initial videos were centered around trivia/fun facts curated by the English Wikipedia Did You Know? (DYK) community through short/simple remixed fun facts. Key Outcome: We learned that the AI tools available today are not able to reliably generate high-quality videos from arbitrary Wikimedia content without human input and oversight. Thus, this approach works well for generating short (under one-minute-long) videos for external platforms like TikTok and Instagram, but without substantially more human curation, generated videos occasionally present incomplete, confusing, or erroneous information.

Does remixing existing Wikimedia content into short videos on popular platforms like TikTok help us bring free knowledge to new audiences (i.e., younger audiences who we know are less likely to visit Wikimedia projects directly) while also celebrating content created by our communities?

Our aim in this experiment was to investigate how Wikipedia can reach young audiences and provide them the opportunity to learn from Wikimedia projects, as well as a path to become Wikimedia editors. Key Outcome: Over the course of October to January, we attained a total of 1.87M impressions and 1.51M in reach from TikTok, Instagram, and YouTube. 1.46M impressions can be traced to Instagram Reels, 373.65K impressions can be traced to TikTok, and 36.35K impressions can be traced to YouTube Shorts. Instagram and TikTok thus stand out as the two more successful platforms in our experiment.

On TikTok, a platform that Wikipedia previously did not have a presence on, we gained 5.2K followers as of January 2025, exceeding our goal of 1K followers by the end of Q3 of FY24-25 (March 2025). For Instagram and YouTube–platforms where we had pre-existing accounts on (and where we published other content unrelated to this experiment)–we gained 14.2K and 2.2K followers, respectively. The following graph represents the trackable gender breakdown of our audience across all platforms. Overall, our audience leans towards male-identifying users (57.6%) over female-identifying users (20.7%). A notable portion of our audience (21.7%) remains unknown in gender. We know that Wikipedia readers also skew more male (63% as of the [Gaps Index/Measurement/Readers Survey 2023|2023 Global Readers Survey]), a proportion that has remained stable in reader surveys year to year over the past decade), but that young Wikipedia readers report much lower usage of social media than the general population of young people, so we suspect that we may be reaching a mix of young people who are already familiar with and use Wikipedia and some who do not.

What kinds of encyclopedic-based content do younger audiences want to see and engage with on short video platforms? E.g., evergreen knowledge content, trivia/fun facts, topical content/current events, other? What kind of content presentations are most successful on the platform? E.g., short image + facts, longer-form deep dive on topics?

Our successful videos generally had to do with unique, regional-based facts and trivia. For example, our most successful video across platforms was about flat-roofed pubs in the UK. This video attained more than 229K views on TikTok. Another noteworthy example is our video on the Elver Eating World Championship, which attained more than 45K views on TikTok. Both of these topics were DYKs created and nominated by the same Wikipedia user (User:Dumelow). With their guidance, we also tested videos made from other DYKs they had worked on. We did not see the same level of virility for these videos, even though they also covered regionally-relevant topics, used higher quality images, and covered subjects that were in the news at the time (e.g., Sycamore Gap Tree). We suspect that uniqueness of subject matter and subjects that encourage discussion/debate are important factors in success on these platforms, but more testing is needed to see if creating content on trending topics is also effective. We began testing longer (1-minute or slightly longer) hand-created explainer videos in January but do not yet have data on how these perform relative to the shorter generated videos. We will update this section when those results are available. We are now inviting community members who are already creating content on these channels to participate in our accounts and will have updates on community-created content later this year.

Can bringing our content to these platforms with clear attribution improve younger audiences’ awareness and usage of our projects?

We are currently awaiting additional user research data to help us understand this and will update this section with some preliminary findings. Early signs suggest that videos on TikTok may be boosting pageviews on Wikipedia, but more data is needed to confirm this. We will update this section when those results are available.

Can we encourage other content creators on TikTok and other short video apps to attribute properly when reusing facts and images from Wikimedia projects?

This question presents a learning opportunity that we have not yet capitalized on.

What metrics did we track?

Key metrics

1. Views on content
2. Viewer demographics (age, gender, region)
3. Engagement on content (likes, comments, shares, saves)
4. Engagement from broader TikTok community (Duet, Stitch, mentions)
5. Qualitative feedback (via surveys, Brand Health Tracker, etc.)

We will be comparing reach, engagement, and demographic data across different short video platforms.

How to stay updated on insights from this experiment

As usual, we will be sharing updates on this and other Future Audiences experiments during our monthly open community calls. Please sign up here if you’d like to be notified for upcoming calls.

If you have any further questions/inputs please get in touch on the talk page!