We are in the middle of a significant rebalancing in society’s relationship with social media.

As Ofcom noted last month, around half of adult social media users ( 49%) in the UK now actively post, share or comment online, down from 61% in 2024. People are changing their behaviour in response to a knot of concerns around privacy, trust and the impact of online harms on our physical, emotional and mental wellbeing.

I have long believed that this kind of rebalancing is both desirable and inevitable, especially as a father of three Gen Alpha children who will only ever know a world that exists simultaneously across physical and digital realities, with all the opportunity and risk that entails.

These behavioural changes will reshape how we understand the role and utility of social media. But they matter for research, too.

For social listening sceptics, Ofcom’s statistics will inevitably reinforce the view that these data sources are not representative; that they only capture the loudest voices, vested interests or cultural extremes.

I think that is a huge miscalculation.

Of course, all research methods have limitations in their ability to be truly representative. But equating a rise in passive social media use with a decline in the value of social listening misunderstands both the nuance of this cultural shift and the full potential of social listening as a tool for better strategy.

Firstly, how people engage with online news, creators and cultural influencers – whether actively or passively – is critical to understanding the emerging cultural shifts that affect every brand, in every category. This is especially true in a volatile world, where important clues about how the future might materialise are often buried in the corners of internet chat rooms or coded into Instagram reels. In that context, pure scale matters far less than technically skilled search and nuanced analysis.

Secondly, while users may be shifting towards more passive usage, Ofcom’s research also suggests that those who do post are doing so more mindfully – that chimes with what we have seen in our own work in recent years.

Yes, we have lost some volume, but much of what has disappeared was low-quality noise. What remains, in the hands of skilled analysts, can be a higher-quality data set, one that is more closely aligned with authentic audience perception and, in many cases, offers a more natural syntax for consumer needs than people are able to articulate in a focus group or survey. Fewer pictures of cats. More genuine conversations about things that matter.

For a discipline that has too often pitched itself as an AI-powered budget hack rather than a source of creative inspiration, changing online behaviour offers an important inflection point for social listening.

With half of adults still actively posting, there is still more than enough scale. Now is the moment to refocus on data quality, veracity and the still under-realised potential of social listening as a source of differentiated points of view.

Really unlocking the value of social listening means shifting the focus away from big data dashboards and AI automation, and towards something more grounded in human craft and multifaceted analysis.  

Tool choices need to be intentional. We need to curate source materials with a clear understanding of where audiences actually live online. Most social listening platforms, for example, massively over-index results from X because it has the most open API. In our own work, we often choose smaller sample sizes from curated forums, chatrooms and Reddit because they get us closer to genuine audience sentiment. It’s an important distinction, especially given Reddit use has increased by 88% in the proportion of UK internet users in two years, and the platform has now overtaken TikTok as the UK’s fourth most visited social media service.

Most critically, analysis cannot be left to AI alone. We need to build analytic models around a true immersion in the relevant business challenge and match the right model to the right objective. This means social listening plus blended methodological thinking to unlock meaning beyond just identifying conversational trends.

This can include applying models like PESTEL (Political, Economic, Social, Technological, Environmental and Legal) to understand the impact of external cultural influences on audience motivations as part of segmentations, habit loops and semiotics as foundation for interpretation of unstructured data in innovation projects and scenario planning.

Ultimately, we can hardcode those models, or build agents to operate them, but the thinking has to be human, as does the crafting of the point of view or the ‘so what?’ which forms the nexus of the value we create.

As cultural attitudes continue to shift, the debate around the value of social listening could easily harden the scepticism of its detractors.

But it could also become the moment the industry matures: moving beyond volume and automation, towards deeper insight through a better balance of craft and code.

Yes, posting is down. But there has never been more gold buried in the hills. We just need to dig deeper.

Mike Norris is head of Truth Intelligence at Truth Consulting