What 212,904 video posts tell publishers

The conversations show how newsrooms plan, produce and judge video. Beyond these qualitative insights, we wanted to include data as well: what publishers actually post and how audiences respond. So we partnered with Sotrender to paint a bigger picture. Sotrender is a social media analytics company; brands, agencies and media companies use its platform to report on and benchmark their performance on social networks.

For this report, Sotrender supplied data on video posts published in May-July 2026 by 100 established news brands in 15 European markets, on Facebook, Instagram, TikTok and YouTube. The analysis and interpretation are ours. Below, we explain what the data covers and how we read it, before the findings.

What this data can and cannot tell us

The evidence. We chose 100 news brands in 15 European markets and their main accounts. Sotrender supplied data for 348 of those accounts; we use 346 (two Instagram accounts had no usable view counts). The data comprises totals covering 212,904 video posts published in May-July 2026, and lists of each account's most-viewed videos, which we call top-post lists. We analysed the data; the findings and recommendations are ours. The sample does not represent the whole European news market. It includes video of different lengths, not only short-form video or videos confirmed as YouTube Shorts.

What the counts mean. A post is a single item published on a platform; one production can appear in several posts. View counts are the figures each platform reports through its API, as collected by Sotrender in August 2026. Platforms define a view differently, and counts can include repeat plays; they do not identify unique viewers or completed views. Counts are totals at the time of collection, not views received within May-July. We compare views only within each platform.

Means and medians. “Mean views per published video” is total video views divided by the number of video posts. A few hits can raise it sharply, so it does not describe a typical video. “Median” means the middle value when the values are sorted.

What is missing. We cannot separate the effect of editorial choices from that of paid promotion, recommendation algorithms, follower numbers, crossposting or access to agency and TV footage. Instagram view counts can include views of the same Reel on Facebook: the figure Sotrender obtained from Meta's API combines them, so Facebook and Instagram views should not be added together. We have no data on production costs, staff time, watch time, revenue or conversions. Comparisons identify examples to study; they do not prove what caused an advantage.

How to use the report. Use the findings to choose what to examine or test in your newsroom. The recommendations below are suggested actions; we have not measured their business results. Detailed tables, links to the posts we cite and the full list of brands are in the Supporting evidence section.

On Facebook, video was a small share of posts but usually earned a stronger response. For the median Facebook account in our sample, video made up 8.7% of posts. Yet video earned more interactions per post than non-video posts at 75 of the 79 accounts that published enough of both to compare. For newsrooms focused on articles, this is a reason to test video earlier in story planning. The data has no link clicks, so it does not show the traffic value of posts that link to articles.

High publishing volume is not the only route to high total views. On Instagram, Brut and la Repubblica each recorded about 2.9 billion video views. Brut published 1,297 videos; la Repubblica published 2,803. They are useful examples of different publishing patterns, not proof that one newsroom uses its resources more efficiently.

A single viral video usually accounted for only a small share of an account's total views. Among accounts publishing at least 100 videos, the median share of quarterly views from the most-viewed post in Sotrender's top-post list was below 5% on every platform. Taken together, the listed posts (roughly 1-3% of an account's videos) accounted for 18-28% of its views; both figures are medians across accounts. Study the hits, but assess a series across all its episodes.

The most-viewed video is not always the right model to follow. For 53 of 87 eligible TikTok accounts, the most-shared post in the top-post list was not the most-viewed one. On YouTube, the best-performing video length depended on the measure. Choose the purpose of a video first, then decide how to judge it.

Facebook shows the widest gap between how rarely video is posted and how strongly audiences respond to it. For the median Facebook account, only 8.7% of posts were videos. Seven of the 92 Facebook accounts published more video posts than non-video posts.

For the format comparison, we used accounts with at least 20 posts of each type. Video earned more interactions per post for 75 of 79 Facebook accounts. At the median account, video posts earned 3.31 times as many interactions per post as non-video posts. The result also held at stricter thresholds, up to a minimum of 100 posts of each type. The same was true without shares, for which the dataset shows none on non-video posts at 27 of the 79 accounts: counting only likes and comments, video earned more per post for 75 accounts, with a median advantage of 3.23 times.

Instagram was different. For the median Instagram account, video already made up 52.3% of posts. It earned more interactions per post for 54 of 81 eligible accounts, with a smaller median advantage of 1.22 times.

Table 1. May-July 2026. Video advantage = mean interactions per video post ÷ mean interactions per non-video post, within the same account. Median video share is calculated across all accounts on each platform (92 on Facebook, 97 on Instagram and 88 on TikTok); the other columns use accounts with at least 20 posts of each type. YouTube is not shown because the dataset records only video posts for it. The interactions counted differ by platform: Facebook includes likes, comments and shares; Instagram, likes and comments; TikTok, likes, comments, shares and saves. Compare formats within each platform, and keep these differences in mind before comparing the size of the advantage across platforms.

We also compared the 68 brands that met the comparison rule on both Facebook and Instagram. Facebook's video advantage was larger for 58 of them, and still for 58 when we removed Facebook shares from the calculation. So Facebook's larger advantage holds when the same publishers are compared on both platforms. The detailed checks are in A2 of the supporting tables.

For a newsroom built around articles, a practical step is to include video in story planning instead of asking for a clip after the article is finished. Pick a type of story you cover regularly, where footage, a demonstration or a short explanation would help the audience. Test whether video adds value there. The data does not point to an ideal share of video posts.

On Instagram and TikTok, ask a different question

Accounts that do well with video can also have strong non-video posts. Tagesschau recorded mean interactions of about 25,600 per video post and 70,300 per non-video post on Instagram. BBC News recorded about 20,100 and 38,400 respectively. Dropping those formats simply because video gets many views would risk losing a large part of the audience response.

The data does not separate photos from carousels. Before replacing a successful series with Reels, identify what your audience values in the current format. In the report's interviews, Daria Volpe describes how VD News is selective on Instagram, mixing videos, images and carousels, while testing more formats on TikTok. The interviews describe individual newsrooms' experience; they are not a survey of the accounts in this dataset.

TikTok is already almost entirely video: the median share was 98.8%. Only 29 accounts had enough non-video posts for our comparison. The main question there is what the videos should achieve, rather than how to increase video's share.

In practice: look for stories that deserve video on Facebook; keep the non-video formats that work on Instagram; on TikTok, be clearer about what each video is for. A single publishing rule will not serve all three.

Similar total views can come from very different numbers of posts. Brut and la Repubblica make this clear on Instagram. La Repubblica recorded 2.903 billion video views from 2,803 posts. Brut recorded 2.895 billion from 1,297, fewer than half as many posts.

On TikTok, Daily Mail / MailOnline recorded about 2.38 billion views from 1,068 videos. La Repubblica recorded 757 million from 3,123. Both totals are large, but they come from very different combinations of volume and mean views per video.

Table 2. Two separate comparisons within platforms, May-July 2026. Mean views per published video = total video views ÷ video posts. It does not describe a typical video or measure production efficiency.

Brut's result raises a useful question for a head of video: how does this publisher choose and present stories? It does not prove that posting less causes better results. Brut's audience, distribution (including any paid promotion) and production resources may differ from la Repubblica's.

Choose examples that fit the question you are asking

Rankings change when the measure changes. On Facebook, only three of the top ten accounts by total views were also in the top ten by mean views per video among accounts publishing at least 100 videos. A team interested in total views and a team reviewing how its own posts perform will need different examples.

Some useful examples come from accounts that publish far fewer videos than the biggest publishers. RTL info recorded mean views of about 431,000 across 146 Facebook videos. HLN recorded about 473,000 across 222, and Zeit im Bild about 397,000 across 185. They combine lower publishing volume with high mean views.

HLN, RTL info, Zeit im Bild and BBC News also ranked in the top 25% for mean views per video on Facebook in each of the three months. All published at least 20 videos in each month. Their results therefore do not rest on a single quarterly average, which makes them worth a closer look. The check shows that they ranked highly every month; it does not show that their results improved from May to July.

Use total views to find examples of scale. Use mean views per video to compare how a publisher's videos perform on average. Then examine the actual videos, including weaker ones. Choose peers with comparable audiences, subjects and resources; a Europe-wide ranking is only a starting point.

More posts need a clear purpose

Across the sample, accounts publishing more video generally recorded more total views. The link with mean views per video was much weaker: a rank correlation of 0.04 on Facebook and 0.37-0.46 on the other platforms, against 0.81-0.89 for total views. Yet some accounts combined high volume with high mean views. The data supports neither “publish more at any cost” nor “fewer videos always perform better”.

The monthly comparisons cannot settle whether publishing more helps. All posts were measured in August, so May's posts had longer to collect views than July's. We therefore report those comparisons only in A5 of the supporting tables.

In practice: before publishing more, define what the extra posts should achieve (useful coverage, a wider audience or a specific action from the audience). Judge comparable videos after the same number of days and record the extra work. A bigger total is worthwhile only if it serves a purpose that justifies that work.