Leah asks, “Is social listening useful beyond Twitter?”
This is an interesting and important question. Social listening platforms vastly overindex on Twitter not because it’s a great social network, but because it’s the last major social network that shares its data so freely. Every other platform has locked down its data to third parties, and so every influencer marketing software package and social listening software package is over-reliant on Twitter.
There are plenty of other sources of data about audience conversations; more advanced platforms like Talkwalker bring those data sources in as a way to compensate for major networks like Facebook, Instagram, LinkedIn, and Tiktok not providing data. Social listening is definitely useful if your audience participates on those platforms, platforms like Quora, YouTube, Twitch, Pinterest, etc.
The challenge many marketers run into with any above-board social listening tools is that today’s tools are excluded from gathering data from the largest platforms – Facebook’s family of apps, LinkedIn, and Tiktok as the leading examples. And no tools of any kind exist for extracting information from velvet rope social media communities, those like Discord and Slack, where conversations are completely opaque to the outside world.
That’s not the major problem with social listening tools, however. Those problems are obvious, and social listening tools still provide useful qualitative data; we can surface new and interesting ideas and conversation snippets from what we do have visibility into.
The Big Social Listening Problem: Representation
Here’s the much bigger problem with social listening tools, a problem that no software has ever solved and probably will never solve because it’s a statistical issue that we simply don’t have enough data to solve: representation.
Let’s look at a couple of examples. Let’s say you’re doing social listening on a topic, and your software can monitor Twitter, Quora, blogs, etc. You extract a big pile of data, garner some useful content, and get some nice charts.
Now, the question is, is that representative of the audience you care about? Here’s a diagram of people we could monitor on social and our ideal audience:
If this is the case, then our social media data is a good fit for representing who our audience is.
However, what if this is the case?
Now you have a real problem. The percentage of people who are loud about a topic you care about on social media largely do not represent the people you want to pay the most attention to.
For example, suppose you’re a coffeeshop and you’re listening for conversations about you. Everyone’s complaining about your building location, it’s ugly, it’s inconvenient, the parking is a pain. What decision should you make based on all that feedback? Should you invest in making changes to your location?
Now suppose almost everyone complaining doesn’t actually go to your coffeeshop. Instead, it’s a social media campaign organized by a local neighborhood activist group that objects to your company’s presence in the community. At best, any changes you make would be a waste of money, because you’re not really listening to your actual customers. At worst, you might alienate customers with your changes.
How do you solve for this problem? How do you know whether the social listening you’re doing is valuable or not? There is an easy and inexpensive way to solve this challenge! Here’s an example I received recently from Lab Alley:
That’s it. It’s easy and straightforward. ASK YOUR CUSTOMERS. When they reply, if no one says any social media channels you’re doing social listening on, then there’s a good chance social listening isn’t going to be a super representative tool. For example, if the only social network people respond with is LinkedIn, then there’s no social listening tool on the market that will deliver useful data to you.
Social listening is useful, but beware the representation problem. Make sure the data you’re collecting is from and about the people you should care about, and it will serve its purpose.
You might also enjoy:
- Understand the Meaning of Metrics
- How To Set Your Consulting Billing Rates and Fees
- It's Okay to Not Be Okay Right Now
- Transformer les personnes, les processus et la technologie - Christopher S. Penn - Conférencier principal sur la science des données marketing
- Marketing Data Science: Introduction to Data Blending
Want to read more like this from Christopher Penn? Get updates here:
Get your copy of AI For Marketers