Strategy


4 Ways to Use Insights from Your Social Media Data to Drive Brand Strategy

Share This

SUBSCRIBE TO MONTHLY POST RECAP

SVP, DIGITAL, MEDIA & ANALYTICS

Todd LaBeau

Quotes John Hughes movies 3x a day. Every. Day.

From Marriott to Southwest Airlines, it can seem like everybody is trying to figure how to use social media data. So much so that making sense of social data is predicted to become an almost 10-Billion-dollar industry by 2022.

The raw truth: If you know how to use them, social media platforms can provide you with all the data you need to help you find valuable insights for your brand—for free. The individual posts and user information can be pulled directly from Facebook, Twitter or just about any other social media site through APIs¹. The data include text of the actual post, the time it was posted, the location of the user when it was posted, and other valuable information on users’ profiles.

The problem? It’s a lot of data. And most of the solutions for analyzing all this data are still clunky, expensive and frankly, not very intuitive to use. We went a different route and found our own solution by tapping into the same data processing tools as our neighors down the street at the University of Wisconsin-Madison. It means getting our hands dirty with a lot of different tools and digging through large quantities of data, but hidden inside the firehose of raw data are gems that you can use to punch up your approach to your consumer.

So what exactly can you learn from this social media data? Here are a few ways we use raw data:

1. Track sentiment of topics

Knowing how people are reacting to your products in real-time can be invaluable to your long-term success. Maybe you just launched a new ad campaign, or your brand was recently in the news. Detecting trends and sudden shifts in sentiment can help your team identify whether you should amplify a message that is working or to try something new.

For example, when Grupo Bimbo, a Mexican snack producer, noticed their online sentiment spiked in Mexico after they launched a new product in the United States, they responded by introducing the new product in Mexico as well, where it was an unexpected success.

You can also track aggregate sentiment to compare brands to one another. Just to test this out, we experimented by pulling tweets about cereal brands. (And yes, that’s our idea of fun!) We then aggregated the tweets and used our machine-learning algorithms to measure which brands have the most volume and highest sentiment in online conversations. The results were interesting.

This type of information could be used to inform future small-bet tests or strategies. If you’re like Cinnamon Toast Crunch and have a ton of positive conversation about your brand, it would be advantageous for you to create easily shareable content for your fans.

What if your brand is like Frosted Mini-Wheats and isn’t talked about very much, and when it’s discussed, it’s less positive than the other brands? In this case, you could get the conversation going with sponsored posts that show off the positive side of your brand.

2. Analyze how volume changes over time

With raw social media data, you can identify the exact times people are talking about your products. This understanding can give you insights as to how people are using your products and what times you can enter their online conversations.

You can use this information to find what is already working or look for new opportunities. Facebook content strategies rely heavily upon finding the right time to post your content and this type of social media data can help you understand exactly when people are most primed to engage with your content.

For example, here is a graph that shows the distribution of tweets about three different beers by day of the week. Notice how the conversation about each beer is unique.

  • Leinenkugel’s Summer Shandy gets a slight bump on Wednesdays and peaks on Fridays.
  • New Glarus’s Spotted Cow is most talked about on Saturdays.
  • These two contrast with Bud Light, which is more evenly talked about across the week. Notably, its distribution of online conversation on Thursdays is an outlier compared to the other two beers. Not surprising; Bud Light’s Thursday night football sponsorship/activity is likely paying off.

The battle between beers is well documented. And so, Leinies might want to consider pre-empting Bud Light and ramping up its paid Facebook strategy on Wednesdays and Thursdays.

Alternatively, within Wisconsin where Spotted Cow is exclusively sold, New Glarus could leverage the insight that the distribution of its online conversation is bunched on Fridays and Saturdays by focusing its marketing to get more Wisconsinites to drink more Spotted Cow during the week.

3. Discover more about the people talking about your products

Social media data mining isn’t all about what people are saying. Another way to use social media data is to find out more about the people who are talking about your brand or category.

For example, you could pull the location, language and Twitter bios of people who tweet about any topic. Maybe your company notices that a lot of people who are tweeting about your product list Spanish as their default language in their profile. If you’ve never done Spanish advertising before, this might be a good clue that it’s time to start.

Or maybe you’re a protein bar producer and you discover that people who are tweeting about your products often say in their bios that they’re runners, but few say that they’re weight lifters. This insight could tell you what types of people already are talking about your products and the types of people who are currently not.

You can also learn more about the people who were talking about an event or hashtag campaign.

For example, we collected tweets that mention “pumpkin spice lattes.” We then scraped the Twitter bios of the people who sent those tweets and ran them through some custom algorithms to find the different groups of people who were using this phrase.

We discovered that there were nine distinctive groups of people talking about everyone’s favorite fall drink.

Notice anything interesting that could apply to your brand or products?

While this example includes a product, you could also search for people who are using hashtags or performing an action, like, #ShareACoke or “drinking orange juice.”

You can leverage this information by better customizing your social media content based on the segments of people that are already using your hashtag or talking about your brand or category.

4. Find influencers in your brand category

It’s easy to find people who get a lot of retweets and likes on posts. But you can also use raw social media data to pull lists of people’s followers.

For example, you can use people’s public tweets to model the types of people you want to target. You can then pull the list of people that they follow and identify their top influencers. Sometimes the most influential people aren’t the ones with the most followers, they’re the folks who are followed by the exact people that you want to reach.

These top-tier influencers might be unavailable or inappropriate for your brand, but we can use the data to dig in even deeper and find potentially otherwise overlooked influencers. One way could be to take those top-tier influencers and pull a list of people who they follow. Then by using network analysis, you can use this list to effectively find the influencers of your influencers who might be a better match for your brand.

A great example can be found in the cannabis industry. PR is a primary communications challenge in cannabis due to the limitations and restrictions of digital marketing of marijuana and associated products, and influencer marketing is a smart strategic approach. With social data, you could start by finding lesser-known influencers. We would start by making a list of top-tier influencers, discovering lesser-known people they follow and calculating their influence on the online network.

For example, we created a list of top-tier cannabis influencers. We then scraped a list of people who those top-tier influencers follow on Twitter. By using network statistics, we discovered the following folks:

  • A writer-producer in LA with just more than 2,000 followers on Twitter
  • A drug policy reformer and social worker with more than 5,000 followers
  • A medical marijuana cultivator who’s just launched a vegan organic nutrition supplement who has less than 5,000 Twitter followers

These are influencers who were not on our initial list of top-tier influencers, but they still hold significant influence within the online cannabis conversation.

Furthermore, if there’s an influencer you desperately want to connect with, but you are having difficulty reaching them, you can use network analysis find sub-groups of the larger conversation.

Let’s say you really wanted to get on a particular influencer’s radar; you could use a clustering algorithm to find influencers within their segmented online network. These other influencers could allow you to indirectly reach those top-tier influencers.

Wrapping it up

Jumping into the world of raw social media data can be intimidating and can easily make you feel like you’re looking for a needle in a haystack. Here’s the key: Remember to use your data to be explanatory, not exploratory.

Strive to save yourself time and find concrete insights. Start your analysis with clearly defined questions.

Use this list as a place to start:

  • Are people talking positively or negatively about my products?
  • What time of day do people talk about my category?
  • What are the top-five words that people who are tweeting about my brand use in their Twitter bios?
  • Who are the influencers in my brand’s category?

You never know what you might uncover—and where it could take your brand.


¹ Twitter API accessed via retweet R package