Over the past decade, how have you seen social listening evolve from a
tactical reporting function into a strategic decision-making tool within large
organizations?
Honestly, it's been one of the most satisfying things to watch over the last decade. When I started, social listening was the thing you did after the crisis happened. It was entirely reactive and in retrospect.
Now the best teams are using it to inform product decisions, media strategy, and brand positioning (voice, tonality, audience, where to play, etc.) well upstream of any in-market activations or campaigns. In my mind, this began when practitioners stopped asking "what did people say?" and started asking "what should we do next?".
This is the evolution of descriptive to prescriptive analytics.
Many orgs stopped treating social data like a news ticker and started treating it like an insights engine. And it’s very clear today which brands those are, whose marketing is rooted in consumer insight and whose isn’t.
Here's the thing: “interesting” doesn't pay the bills. It’s what I call “FYI” as opposed to “actionable”, things that are unique but aren’t valuable enough to force an action or reaction.
The teams that drive impact know who they're talking to before they pull a single query. They've typically got a stakeholder who's already bought in, a data-literate culture, and have full trust in social data and the people analyzing it.
They’re able to translate “here's what we found on Reddit” into “here's what this means for Q3 creative”. Audience ABC, persona DEF, across one or two channels, X% organic vs. Y% paid, and rooted in human truth XYZ (taken from social listening).
The teams stuck at “interesting” are still reporting on what happened. Impact teams are answering questions nobody thought to ask yet. That's the gap… it boils down to skill and the ability to prescribe actions. It’s one thing to be good at analysis, it’s another to turn it into a business solution.
Traditionally these sit separate from each other, with slightly different metrics, and I don't disagree with that. You can action off of both independently.
But they fundamentally represent two different forms of value creation: one from things you control (owned), one from things you can't (earned).
I place higher value on earned simply because it's a greater representation of resonance and brand love, and you simply can't sell things to people who don't know or care about your brand. Consumers defer to other people's opinions... and that's never been more readily available than it is right now.
Owned complements this. It's your identity, your story, your narrative. Things you control.
The metrics are very similar across both, consumers and creators log into the same Instagram brands do, and use the same measures to determine success. But each carries different meaning, and depending on the analyst, different weights. I weigh earned 60/40 over owned.
I still prioritize based on volume. If more people are active on one app over another, you want to be there more. Call it mirroring.
The exception is TikTok, where available data significantly understates actual conversation volume. So I weight it higher than the numbers suggest.
Reddit is having a moment, and social listening supports the case to be there... but just because the opportunity exists doesn't mean brands will act on it. Organic is a hard sell when you can't compute ROI the way paid media can.
I'd also say, prioritize where the broader brand is already playing. That should always be a consideration.
Most of the major SaaS platforms have integrated their own AI built on top of the big LLMs, so you see it baked into everything: search creation, widget and data summarization, report generation, and more.
I'm a big fan of using it for text summarization, which is very controversial right now. People claim these models hallucinate, which is true, but I've had very few of those experiences in my roles and datasets.
As someone who ran this for a massive portfolio, I needed a computer to read mentions faster than I could. There's just no way around it... at a certain scale you're compromising on accuracy and time. We can analyze mentions in seconds that would normally take hours.
And when you're forced to do the work of five people, you quickly become an advocate for more accurate NLP in these models. We are already here, this isn't the future..
As AI generated content increases, do you think social listening datasets risk becoming less reliable? How should analysts adapt?
I think the acceleration of AI-slop makes people more hesitant, but the fundamental benefit of social listening is that you cut through the noise and avoid everything around you.
I've said the same thing about Twitter/X. A lot of people root against the app because of Elon, but at Haleon it was still a valuable source of consumer insight... because we cut through the noise of everything around it. That never changed.
GenAI content is the same thing. It inundates the feed and frustrates users, who every day get better at identifying it. If anything, I think this is giving people a greater appreciation for original content and unique POVs.
The tools will catch up, they always do. In the meantime, my advice is to expose yourself to more content and to the models themselves, because with enough practice you'll be able to pick them out.
I think data collection will become a lot more autonomous, in both topic creation and the elimination of Boolean queries... at least the need to manage them manually on the front end.
And the breadth of where conversation lives keeps expanding. We're already seeing a resurgence in Substack and Medium, people taking ownership over their content and expertise. Long-form is here to stay, and wherever conversation goes, social listening follows.
I actually think many of the fundamentals will remain the same. It's the velocity and volume of data around us that is growing exponentially. The analytics and data mining methods are well established, and will become increasingly useful.
The world is investing more in social intelligence, not less.
Bonus question: What's one widely accepted "best practice" in social listening that you believe is outdated or no longer effective?
I don't know if it's necessarily a best practice, but text-based analysis as the default way we understand conversation is becoming less and less reliable. It made sense when that's where conversation lived. It doesn't anymore.
Video is where culture moves now, and most of our tools and skills were built for a different era. So if you're only listening to what people type, you're missing most of what they're actually saying.
Years ago I remember saying... if we could get AI to watch videos and transcribe them into text, we'd make video insight accessible at scale. Description, context, semiotics, all of it. We're getting there, but most practitioners are still leading with text-era skills by default, and that's the habit I'd challenge. Starting with myself.
