Analyse van Bernard Cools: Sources of information in Belgium

Two years ago, Bernard Cools (Chief Intelligence Officer at Space) published his analysis of how Belgians perceive, consume or avoid news. Since then, new data has been published. He strongly recommends the synthesis made by Xavier Degraux based on the most recent information provided by the Digital News Report (DNR). For now, Bernard Cools wondered whether what he had observed and published was still valid. In addition, the DNR has introduced new features.

Main learnings 2026

In many ways, the conclusions drawn earlier still apply: from 2016 to 2026 (the latest data), the number of information tends to decrease, print news continues to decline, the “news diet” of Belgians keeps being screen-based and consisting of a majority of free channels. Payment of digital news still involves a minority of users (14%). Social media are definitely the main source of news to significant proportions of the Belgian adult population.

Trust once again appears to have a complex relationship with usage. While print sources are much trusted, they are not read in proportion and we find the exact opposite with social media or digital news content: distrust does not prevent many people from using them as usual news sources.

New sources have appeared in the latest available data. AI chatbots and content creators were introduced in the DNR  data, respectively in 2025 and 2026. Chatbots remain a marginal source, used by less than 7% of Belgian respondents. Although trust in chatbots is limited (±11%), it exceeds the proportion of users, showing some development potentials.

As mentioned another new source introduced in the 2026 DNR questionnaire: individual creators or influencers. Not less than 20% of respondents declare having used such a proposal in the last week. The questionnaire makes a distinction between creators mainly focused on news and others. The former appeal more to French-speaking respondents (18%) than to the Flemish (8%) while the latter who “mainly post about other topics but sometimes talk about the news” are almost equally used by the two linguistic communities of Belgium. Interestingly these creators are seen as more entertaining and easy to understand than traditional media, but in the meantime they perform poorly on neutrality, trust and expertise. It seems that news reporting has something to learn from the influencers’ sphere.

Last learning: compared to my previous analysis of news sources, it appears that any political alignment is relatively predictive of more interest in news, including the propensity to pay for it. Respondents who declare no political leaning are also proportionally the most reluctant to news payment and the most likely to avoid any news.

Update: state of demand

As shown in Figure 1, after years of relative stability (2016–2020) and a peak corresponding to the pandemic period, interest in news regularly decreased during subsequent years (2022-2024, 2025 ).

The latest data (2025 in the Flemish part, 2026 in both regions) reports a certain stabilization: more than 1/3 of respondents (36%) of the Digital News Report declared to be interested in news (“actualités” in French, “nieuws” in Dutch). As mentioned in previous publications, “news” may be interpreted differently by each respondent, in the absence of any other explanation. Some editions of the Digital News Report polled respondents about their interests in different thematics, making the meaning of “news” clearer. In 2022, for instance, while only 46% of all respondents declared they were interested in news, another question (“Which of the following types of news, if any, are you interested in?”) revealed that 62% of respondents expressed interest in local news “(e.g. about my region, city or town)” or 57% in international news. Among the 16 topics, the least appreciated was “Entertainment and celebrity news” at 26% (average across all topics: 39%). More crucially only 5% of the 2044 respondents of the survey either declared no interest or did not respond. So it might be assumed that respondents rather react based on their perception of “news” as a general notion, but their actual interest in given topics may be somehow intense.

The question about the interest in politics  is not asked every year by the DNR, but when data on this subject are available, they show a remarkable parallelism with the interest in news, as shown in figure 2.Of course, it is a cliché to repeat that correlation does not mean causation, and it is true that both topics are quite general (some would say “vague”) but “politics” and “news” seem to be notions that somehow relate one another.

The much-debated question about active news avoidance does not appear in every past DNR questionnaire, but is systematically asked since 2022. As shown by figure 3, in the 3 last available years, any avoidance frequency involves not less than 70% of Belgians, with virtually no difference between French- and Dutch-speaking respondents. Around 1/8 report a systematic avoidance behaviour, and a total of  40%  ticked either the ‘often’ or ‘sometimes’ options in 2026.  When available, steadily growing avoidance figures may be linked to the diminishing interest in news. And the magnitude of the group of people claiming they never avoid the news -around 30%- looks reasonably consistent with the proportion of those who say they are interested in news (±36%).

Avoidance profiles shown in Figure 4 hardly differ depending on the declared frequency: relatively moderate as well as frequent avoiders are to be found primarily among non-politically aligned individuals, people living in lower income households, women, and respondents with a medium education level. One may also find proportionally more avoiders in age groups 25 to 44 years. Worth noting that the youngest respondents, under 25 years, seems to be infrequent news avoiders.

Update: sources of information

The available DNR data now span a period of more than 10 years, showing remarkable shifts in some cases. The question on information sources first reviews a list of more than 12 options not mutually exclusive. Podcasts and AI chatbots are recent additions in that list. As shown in figure 5, highly significant evolutions have taken place, even at the most aggregated level, all Belgian respondents.

Although television remains the primary source used to consume news, its penetration has decreased from more than 70% to 46%, from 2016 to 2026. Social media platforms are now (2026)- competitive to the digital content of newspapers, that however seriously decreased in penetration, from 59% to 46% nowadays. The printed press also lost much, both in terms of usage frequency and in place in the ranking. More generally non consumers of the proposed news sources has been multiplied by more than four, from less than 2% in 2016 to more than 8% in 2026. The average number of sources used slipped from 3,5 to 2,7, in spite of the abovementioned addition of two proposals since 2025.

Figure 6 shows that the evolution in terms of non-use and average number of sources has been parallel in both linguistic communities.

The DNR questionnaire also focuses on the preferred source of news for respondents (referred to as “MAIN” in the questionnaire, in capitals) who have indicated at least one channel among the different proposals.

In Figure 7, one can see the evolution of the main sources (only one answer is allowed) as declared by all Belgian respondents. Television still comes first as it did more than 10 years ago, but after a peak at 40% in the pandemic period (2021) it declined around 30%. The digital content of newspapers comes as second with diverse but limited fluctuations in the course of years. Social media used to be #5 in the ranking and climbed to the third place in 2023. Radio news has maintained positions in the meantime. On the contrary of printed newspapers, 3rd most used as preferred channel in 2016, 6th in 2026, with a significant loss in share of consumption. The aggregates at the bottom of the table show the respective shares of media displaying their contents on screens, be it television, computer, tablets or smartphones, the print media, and all free sources. In that case, only (quasi-)certain options have been retained: it consists of audio-visual media including their websites and applications, or social media, but not of the sites or apps of news media, that allows free consultation, but not exclusively. In the course of years, free media sources may fluctuate but their share of main usage has been around 2/3 or more of the total respondents bases for at least 4 consecutive years. Information coming on screens involve more than 80% of respondents, a proportion that has somehow increased from 2016. Print -either newspapers or magazines- has drastically decreased in penetration from 12% in 2016 to less than 6 in 2026.

Figure 8 contains the same information as 7 but for Flemish respondents. It shows a relatively slight decrease in the share of TV news that however keeps being the most frequent option as main source of news. Digital content coming from newspapers also confirms its second place in the ranking with penetration figures that may fluctuate but remain within a limited range. Social media are now (2026) #3 in the ranking with a heavy increase of their penetration from 9% in 2016 to 16% in 2026.  The sites and applications of AV publishers do follow: their usage as main source has significantly increased in the course of time. As 5th the news on radio stations has maintained a modest but relatively stable position within the different options. This stability does not apply to the physical newspapers, 3rd main source of news in 2016 fallen at the 6th rank and a penetration divided by 2, while in the meantime the digital content of newspapers did not grow in proportion. 24 hour news TV is relatively marginal in Flanders, and this may partly due to the lack of channels in the local language. When it comes to the aggregates, the proportion of “screen media” used as main sources of news as well as the usage of free media grew in the course of time, while the print options -including magazines- clearly decreased even if the trend did not always consist of a straight line.

Figure 9 (main sources of information for French-speaking respondents) shows that when it comes to aggregates, screen media and free sources rather maintained their positions in the course of years, as they were already at higher levels than in Flanders in 2016. The penetration of print media heavily decreased from 2016 and 2026, and like in Flanders, one hardly sees transfers from print to digital for magazines and newspapers. In parallel with Flanders, TV news comes as first main source of news, but its prevalence has dramatically lost size: in 2026 the penetration of TV news is 1/3 smaller than it used to be in 2016. Social media platforms are now at all-time high, surpassing the websites and apps from newspapers. Radio news and digital news content of AV companies do relatively well, in positions 5 and 6. Alternative news outlets are much more frequently used as main reference by French-speaking respondents than by their Flemish counterparts, where this option is rather marginal. 24 hour TV news channels are also more developed in the Southern part of Belgium, where French channels like LCI, France 24 or C News may attract some audiences. Worth noting that the inception of a local news television channel, LN24, did not result in a substantial increase in use of such proposals.

Next to the general results, one may find striking differences in the “news diet” and consequently in the sources of information between generations. Over the last 3 years, the first source of news for people under 35 are social platforms. They are evenly matched in the 35-44 age group but then strongly diverge, TV becoming the main reference for respondents above 45 years, as shown in figure 10. Showing this huge difference in terms of news source  in  2024, consultant Deloitte commented: “This divide could influence public opinion, political engagement, and societal understanding as different age groups receive and process information through distinctly different channels.” (Digital Consumer Trends, Belgium 2024) Those remarks are more than ever valid.

New: alternative sources or devices

It has already been mentioned that AI Chatbots were introduced in the DNR questionnaire in 2025 and 2026. With relatively modest usage: 5% of respondents did declare using them as non-exclusive source of news in 2025 and less than 7% in 2026. ±1% of respondents used them as main source in 2026 (with of course, large differences by generation). Figure 11 focuses on questions asked to the small group of chatbot users in 2026. The first theme is about the reasons why people use them and the second focuses on the actual usage. In both cases, the most frequent option is about follow-up questions. Next it is not surprising to see that chatbots are used because they ease the access to news, with summaries, sophisticated search queries or fast entry points.

DNR 2026 also introduced the topic of news creators or influencers on digital platforms. No less than 20% of Belgian respondents said they had consumed such content, with a marked difference between North (16%) and South (26%). The question also made a difference between those “who mainly focus on news” and creators who only incidentally talk about news. In the latter option, Flemish and French-speaking are pretty close in terms of penetration (11% globally, with few differences). The gap is larger between the two parts of Belgium when it comes to influencers exclusively or almost exclusively focused on news. 18% of the French-speaking respondents said they used such option “in the last week”, while only 8% of Flemish said so. Compared to “traditional news brands and outlets” their main qualities that the users see in influencers is first that they are “easy to understand” (48%) and entertaining (47%). They are also seen as more “authentic” (37% agreed) and “relatable” (35%). Conversely respondents see the influencers as less impartial  than the traditional media (30% do not agree with the “impartial” proposal), less trustworthy (close to 26% align with that statement) and less competent (22%).

Partly new: trust vs usage

The European Media & News Survey 2023 confirms what data from a previous survey had revealed: trusted media are not the preferred sources when it comes to usage. The most striking example is shown with the contrast between printed press, second most trusted media source (1/2 of respondents) but only used by less than ¼  of respondents to access news, while the online news is considered trustworthy by only 9% of respondents but largely (40%) consumed as news source. There is also a large gap between trust and usage for social media platforms. In the audio-visual the Media & News questionnaire, TV and radio were taken together, but distinguished between private and public media. 56% of respondents considered public TV and radio trustworthy (actually the highest level reported in the survey) against only 23% for private AV media. As usage data are only provided at media level, without making any difference between private and public, figure 12, shows the average trust value, almost 40%, for TV and radio together. At least when considering these two media channels, we may see some consistency between relatively high levels of trust and widespread usage. So the relation between trust and usage looks a bit complex.

Figure 14 combines answers to questions that appear in every DNR survey: trust in the news as a general concept and trust in specific media personally used by the respondents, together with questions appearing for the first time in 2026 on trust in search engines, AI chatbots and social media when it comes to news. Next to the usual language segmentations, figure 14 also displays the results on a group consisting of those who claimed they have used a news creator or influencer in the last week (see above).

In this more digitally oriented group, content provided by chatbots or social media is more trusted than in the global population, whatever the language spoken. But even among these people, there is a clear trust gap between the “most news” perception, and what is provided by platforms, namely search engines, AI solutions or social media. When it comes to platforms, it is worth noting that the trust gap between French-speaking and Flemish respondents is relatively smaller. Because on items like “most of the news I consume” and “most news”, the differences between both is higher than 20 percentage points ! Another important remark may be made on search engines used by at least 33% of respondents in the past week, but only trusted by 23% of them. Even more striking contrast when it comes to social media: used by 62% of Belgian respondents to the DNR “for finding, reading, watching, sharing or discussing news in the last week” , and considered trustworthy by less than 11%… On the contrary, chatbots are not massively trusted (less than 11%) but in the meantime are seldom used for news purpose: less than 7%, suggesting that in that very case, trust levels are higher than usage levels. Time will tell if that small “excess of trust” does last.

Update: digital payment

The question on payment for digital news content is an all-time classic in the DNR questionnaire. At  approximately 15% on average in the last 3 years, the proportion of news payers is relatively larger than the 12% reported in the two first Belgian editions of the DNR. But it appears clearly that the peak reached in 2022 has been a “one-off”.

All in all, as shown by the historical data in Figure 15, payment for digital news is relatively stagnant in our country. In figure 16, the demographic profiling of those who pay shows uplifts among people with higher educational levels, among those with more comfortable income, men and in general younger people, with a very visible over-index on respondents aged below 25. Political alignment is also a very powerful driver for news payment, while the “no reply” to the question of political leaning drastically under-indexes on that item. When it comes to the way people pay for digital news, it appears that the most frequent case is by means of a subscription. ¼ of those who pay for digital news do not only pay for it, but get it in bundles. As the sum of all answers largely exceeds 100% it seems that given the large period of time covered by the question, different payment solutions may coexist or follow one another.

Stagnation

All in all, the general feeling is that few things have changed since my publication. News use is confirmed in its different aspects. Interest in news certainly does not increase, even the general wording “news” might be challenged. The propensity to payment stagnates at best and consequently maintains or even worsens the economic issue for publishers of paid news, in a context of diminishing advertising revenue. New channels, like chatbots and influencers, do not perform in terms of trust. But trust is confirmed to be a weak factor in usage prediction. The optimistic view on that is “stability”. Given the current status of journalism especially in the South of Belgium, “stagnation” seems more appropriate.