Social audio analytics is nothing new. For years, companies like Microsoft have been analyzing human voice and other audio signals across a variety of verticals like enterprise, healthcare, productivity and smart cities. Some of these applications include analyzing audio from customers calling into the call center, media content analysis, medical diagnostic aids, patient monitoring and audio analysis for public safety.
The other day I was part of a Clubhouse room hosted by Jeremiah Owyang on social audio analytics. It started from a tweet earlier in the day where he had predicted 4 levels of social audio analytics. They are as follows:
- Level 1: Report on room duration & attendance
- Level 2: Sentiment analysis & translation
- Level 3: Analysis word of mouth, who’s an influencer
- Level 4: Predicts tomorrow’s conversations
I responded with an additional level which would measure the impact social audio has on behavior change, or action taken like selling products or capturing leads.
- Level 5: Behavior change, driving sales
I’ve been working in analytics and measurement for several years now so I thought I would classify Jeremiah’s levels a little bit differently, especially as it relates to marketing and communications.
I don’t think I would classify them as levels though. Doing so insinuates that the levels are sequential or linear and I’m not sure they are at this point. I would group them based on common KPIs or categories.
Engagement & Growth
These are the vanity metrics and will track Clubhouse room analytics, engagement, number of participants, numbers moderators, quantity of hands raised, how long people stay in the room, and other types of crowd reactions (type of emojis, +1s, applause). Also in this category would be audience size and audience growth rate.
This isn’t unlike influencer analysis across other platforms. Certainly audience size is a factor but also club membership and room engagement for rooms were they are moderators. There could also be an engagement rate based on number of people in a room and the engagement within that room. That functionality is not quite available in social audio apps like Clubhouse yet and the volume isn’t high enough anyway.
That said, I always like to caution others to not get too excited over the bright and shiny objects of new apps like Clubhouse. While there is a lot of hype around social audio in general, it is just one channel among several others for brand activation and reaching audiences. See my post here on Clubhouse influencers to get more insight into this topic.
Currently recording room content is against Clubhouse terms and conditions unless it’s clear that the room is being recorded. This type of analysis isn’t seamless though as it requires several steps to retrieve and analyze the data. There are software applications that use AI to translate audio data into text/voice (think Siri, Alexa) as well as transcription services that use humans to transcribe content manually.
On the technical side, analyzing sentiment requires deep learning applications like natural language processing, digital signal processing, tagging and generation. Analysis will definitely go beyond just tracking a room’s sentiment as being negative, positive or neutral. With today’s technology like NLP it can track intent, emotion and other factors.
The other day I was in a Clubhouse room and we were talking about audio equipment. It was just four of us and I wouldn’t necessarily say that any of us are influencers. A friend had suggested a mic for better audio quality when using the app. He told me the brand name and model number and while in the room, I went to Amazon, searched for the mic and bought it within 90 seconds.
As of today, it would be impossible to track the impact of word-of-mouth and sales. And in full transparency, it’s not that easy to track business impact like this in other channels as well. It’s been a sore spot for marketing and communications for years. Imagine being able to attribute a $15 million sale of software to a discussion on Clubhouse. That type of insight might be closer than you think.
There’s already software today that tracks car traffic (think Waze or Google Maps) and crowd (people) movement. Typically this type of data is used by law-enforcement or the military. But imagine being able to predict the turn out of the county fair or an amusement park based on conversations happening on social audio apps. Or better yet, imagine being able to predict the premiere of a movie based on the quantity of influencers that you were collaborating with.
The opportunities are endless for social audio analytics. As new social audio apps are coming to market, so are the developers and start ups that building apps to analyze users and their engagement levels.