6 Ways To Address The Limitations Of Social Media Listening

Social media listening is all about tracking relevant conversations but sometimes what’s tracked isn’t as relevant as it initially appears. 

For example, Kasey Moore of the site What’s On Netflix, recently pointed out data from Parrot Analytics in The Hollywood Reporter was almost certainly wrong. They claimed The Fall of The House of Usher was the 5th most in demand show on Netflix during the week of February 16-22, 2024.

Only the TV series was released in October while the singer Usher did the Super Bowl halftime show the week before. Additionally, the social media analytics platform ListenFirst shows by measuring likes, shares, and comments The Fall of the House of Usher was only the 581st most popular Netflix show during the same time period. 

Granted Parrot Analytics measures both search and social media, however it’s pretty obvious interest in the R&B singer was being attributed to a TV show. 

The areas most challenging to social media listening tools

In terms of using social media listening to monitor a TV show or brand, understanding which conversations are relevant is more difficult than you might think. For example “Coke” can be referring to Coca-Cola or cocaine. MedCerts is both an online training platform and what you call an excused absence note from a doctor in the Philippines.    

There’s also the issue of brands dealing with intentionally inauthentic conversations. The four hour version of Zack Snyder's Justice League was greenlit for HBO Max, in part because how often the #ReleaseTheSnyderCut hashtag trended. However, reports commissioned by WarnerMedia and leaked in Rolling Stone found “at least 13% of the accounts that took part in the conversation about the Snyder cut were deemed fake.”

Here are some key tips on how marketers can more effectively use social media listening to measure authentic conversation:

#1. Stop worrying about capturing every single brand mention 

In 2023, there were 47M Tweets bringing up the phrase “Taylor Swift”  the hashtag #taylorSwift or the X handle @TaylorSwift. In the same time period, “Taylor” was mentioned in 110M Tweets.

So how many Tweets on X referred to Taylor Swift total? Even if you went through all 110M Tweets, which is wholly impractical, you still wouldn’t know because there are Tweets where Swift is referred to as “she”, “her” or just uses her image. 

The goal of social listening is to get a representative picture of what’s going on. Understanding absolute scale isn’t possible and it’s more important to focus on the data you are 100% sure about.  

#2. Only focus on social media metrics relevant to you

My 2-year-old daughter's favorite YouTube show is Ms Rachel, an educator who helps toddlers learn how to speak. Trying to understand the show’s fans through social listening would be a complete waste of time because the audience can’t post on social media. 

Figure out what metrics matter to your brand’s ROI and make sure those are measured accurately. Measuring everything is too complicated. 

#3. Measure official social handles when doing a competitive analysis 

Understanding noisy data is hard enough around your brand, around competitors it’s impossible. When trying to make apples to apples comparisons between brands, measure the performance of their official social media channels only.

Even focusing on brand names or exact titles is too complicated. For example, good luck accurately counting mentions of Apple or Stephen King’s It. 

#4. Find surrogate terms to track 

If the assignment is tracking mentions of Apple or It, other than tracking official social handles and hashtags, finding related phrases yields the most accurate data. For Apple, use specific product names, with “It” focus on Stephen King or the names of the actors in the relevant time period. 

#5. Do a sample size content analysis

Take 100 social media posts mentioning your brand, categorize by topic and chart how often the posts are actually irrelevant to your brand. Repeat on each social media platform. This illustrates by percentage how much of social mentions are not really about your brand.

It will also create insights about how to save time, for instance there might be entire Subreddits where mentions can be ignored going forward. 

#6. Identifying bot campaigns requires dedicated investigation

For bad actors looking to manipulate the online conversation around a brand or program, AI is making fake accounts easier than ever. For example, the YouTuber Upper Echelon recently made a video showing how using an autonomous GPT Agent, they were able to spike X mentions for the cartoon Velma for less than a $1. 

Because of the complexity of identifying fake accounts, if you think your brand is a victim of bot activity, you need a special type of social media analytics firm. For example, Graphika handled the #ReleaseTheSnyderCut investigation for WarnerMedia, and regularly tracks Russian disinformation campaigns.   

Conclusion 

Social media analytics companies spend too much money on API access and R&D to intentionally provide bad data. At the same time, when you’re tracking literally trillions of data points, mistakes are going to happen. 

If you work with a social media listening solution and a data point seems off, the best thing you can do is ask a lot of questions. Work with your account manager and strategists to make sure the social listening is optimized for your brand’s specific needs.