Don’t Leave Sports-Fan Sentiment Analysis To A Machine

In his August Media Post Marketing: Sports column, SLRG President Jon Last speaks to the dangers of relying solely on web scrapers to analyze fan/customer sentiment.  The human element inserts the added value benefits of context, essential for effective qualitative story telling.

I’ve learned a lot over my career as a sports marketing researcher.  In fact, a zest for learning and immersion in the world of the respondent remain among the most important qualities we look for in our professional team.

There’s no shortcut for this. Actionable research still comes down to rolling up your sleeves.  But at the same time, with an abundance of accessible digital chatter, sports marketers needn’t go far to find conversations that can render insights. Unfortunately, though, the idealistic belief that all technology is good and that “cheap and simple” are best, often leads to cut corners, rendering misinformation or buried insights.

Machine learning, AI and web scrapers that identify language patterns are not a substitute for the human element of considering the appropriate context.  Non-humans lack the ability to assure that the population having the conversation is appropriately sampled or representative. Technology also fails to probe respondents or apply projective research techniques culled from social psychology, as good qualitative researchers can.  Pulling out the raw emotion and motivations behind one’s comments is even more critical in high-involvement, passion-driven categories like sports.

Rather than the “quick and dirty” default of deploying software that finds the most frequently used words across discussion forums or social media, a human-driven approach recognizes significant nuance and emotional motivations. Simply stated, the size of “Wordle” balloons doesn’t help anyone know how “Frustrated” applies to fan sentiment, only that a lot of comments spoke to some form of frustration.

Good qualitative research is about uncovering and telling the customer’s story.  It delves deeply into the context of emotional connection or disconnection with a brand or concept. It phrases inquiries and probes to get at underlying drivers of perception.

Projective exercises can get past the “grandstanding” or detached bravado that often dominate social media or message boards.

One sports project incorporated role-playing exercises to reveal the nuance behind varied marketing messages to different customer segments.  Supplemented with in-the-moment video capture, studies like this literally surface the voice of the customer and allow us to consider facial expressions and body language for better understanding.

That’s not to say that in-person observational research is the only means to get at underlying motivations or build archetypes that bring customers to life for better messaging or brand assessment.

I remain an advocate of collecting organic social media data.  But proper analysis still requires curation and interpretation of this content.

We often utilize what I call “emotional coding.”  Here, we examine comments or posts to tally frequently expressed themes, rather than just the most-popular phrases that are a limitation of typical sentiment-scraping software.  The researcher must examine what is said and how it’s said, as well as what isn’t said.  The human element also provides the critical ability to insert a foundational grounding in the category or topic being studied. Machines lack that value-added capability.