Ever since social media arrived on the scene, the market has been inundated by data providers claiming they can mine huge volumes of public online comments from blogs, forums and other social media platforms and identify relevant conversation themes which can be used to inform brand strategy. Anyone who’s ever used one of these social listening tools will know that it’s not quite that simple.
Although there are lots of conversations out there, I’ve yet to see a tool that does a halfway decent job of automatically identifying or analysing them – and for good reason. For a start, although Natural Language Processing is accurate enough to be useful for content such as newspaper articles and customer feedback forms, online chatter is a different matter. Exceptions are invariably much more frequent than any rules we could create for categorisation. Among other factors:
Given the challenges involved in designing mechanical processes to address any of these factors, let alone all of them, it’s hardly surprising that automated text analysis usually misses so many relevant findings that it’s arguably not worth using. That’s not to say useful insights can’t be obtained from social media data – on the contrary. But the idea that this can be done effortlessly by an automated tool is wishful thinking and isn’t borne out by any of the technologies currently available.
At TMW, we use a combination of random data sampling, human analysis and close collaboration with brand managers and strategy planners to draw out truly relevant and interesting themes from social media chatter. Rather than analysing online mentions of client brands, which is typically of limited usefulness, we look at broader conversation topics, eg, what people are most concerned about when buying a new type of appliance or car, what they do to prepare for a hot date, or what’s most important to new parents in the context of raising their child. We’re looking for what people care about, what pushes their buttons and what their unmet needs are.
For brands where the product or sector itself generates a lot of advice-sharing – eg, a personal grooming product or smartphone – we can identify things like factors driving consideration or confusion, barriers to purchase, positive and negative triggers, brand and competitor perception...the list goes on. We can also analyse broader topics or areas of strategic relevance – eg, we might be targeting people who attend music festivals and want to understand what makes the experience special for them. Whatever our angle, there’s a good chance there are conversations which can provide useful insights.
In summary, the social media space is teeming with conversations that are potentially useful to marketers. In fact, the things that get people talking online are exactly what we need to listen to – the things they care about which they want to share or express frustration or seek advice about. If our aim is truly to understand what people are thinking, unprompted social media chatter, analysed using a combination of listening tools and human analysis, can tell us a great deal.