How We Built Audience Personas for Crocs Using Social Listening
by Infegy Research Team on November 23, 2021
Holy hole-filled shoes, batman! Crocs are back.
Crocs garnered a 30% increase in social conversation in the past year and grew their revenue 73% from 2020 to 2021. Surprising right? The ugly step-sister of the shoe universe has suddenly, and loudly, taken social media by storm.
But who are the people driving this trend? How can we find out who this brand appeals to and what about their lives and lifestyles makes Crocs -- a once highly-niche shoe product -- fit them so well?
If we want to understand the “why”, we need to understand the consumers who love Crocs and post about them online. Using social listening, we can build out online social personas for the Crocs brand, and you can repeat the process with your own brand too.
Don’t know which segments are core to an online audience? We happen to have a social listening tool, Infegy Atlas, capable of powering audience segmentation and persona building. There are so many possibilities with social listening especially with a flexible and speedy tool at our fingertips.
We were able to use social listening to build 5 brand personas for Crocs, which will be featured in our email series on audience segmentation.
Ahead, we’ll walk you through the steps to explore the unknowns and uncover core audience segments using social listening. We’ll use that data to find out who gave the uniquely positioned Crocs brand a comeback (really who in the world would wear Crocs?!).
Ok, enough snark. Let’s build an audience segment!
Step 1: Identify a research question
First and foremost, it’s crucial to start with a clear research question.
Starting with a clear question will guide our search parameters, help us sort through lots of data points, and ensure our findings are actionable.
Examples of good audience related research questions include:
- Who are new or potential customers? Where and what are they currently talking about?
- How do customers talk about an industry or product? Where are customers talking ?
- What do customers prioritize? What do future or growing segments of customers prioritize?
- Which influencers are most impactful to current or potential customers?
Once you have a foundation for our social listening research, you can begin finding insights about the brand’s audiences.
Step 2: Set general search parameters to define a sample of social data
Social listening provides ample opportunities for researchers to identify patterns in audiences so that you can begin grouping them based on how they represent themselves online.
In this case, the research question is specific to a brand so including online brand mentions in our parameters is a natural starting point. But we also want to apply more context by establishing a timeframe and/or looking for the forces driving the brand conversation.
The Crocs brand regrew in popularity in light of our weird new pandemic times, so setting our search timeframe for this year is logical.
But we don’t want to stop our parameters there. We want to get to the consumers proudly wearing Crocs - not just casually mentioning them or even making one-off jokes about the brand or spouting off a random opinion.
To do this we used key phrases and hashtags associated with behavior -- not just opinions. Honing in on consumers wearing Crocs is key so search terms included phrases like “outfit of the day” and #ootd.
Step 3: Dive into online bios
Alright, we’ve entered in our search to Atlas. It quickly returns thousands of data points -- like in seconds, no need to prebuild queries.
You can research on the fly. The best part is this data is intuitively visualized - no additional cleaning or visualization work is needed.
Your analysts or clients with tight turnaround times are going to love you for cutting out this work!
But, the number of data points and possibilities can be overwhelming. It’s easy to get lost in interesting data that might not help you find your core audience segments. The research team here at Infegy recommends as a best practice first looking at source bio top topics and source bio clusters.
The nature of source bios means consumers are classifying themselves. Their choices for how they describe themselves tells us what is most important to them and how they identify with the world. Source bios are a highly accurate metric when it comes to personal identity and useful when determining different persona breakouts.
Streetwear sneakerheads breakout as a cluster in the Crocs source bios:
Often times the first clustering result you see will be useful.
However, depending on your search it might be useful to make the clusters more expansive or more exclusive. A unique Infegy Atlas feature is the ability to hyper tune the clustering “model”.
Essentially you can choose if you see more or less topics, how easily topics link, and how easily topics cluster. These are super easy adjustments made with a few clicks, so no advanced stat skills are needed.
*Tip: when clusters yield an overwhelming number of topics using “cntrl F” and searching for specific keywords can help get you started with finding relevant clusters.
In addition to clusters, Source bio topic frequency can be a valuable data source for understanding top persona descriptors.
Frequency is a less granular look than clusters. It's a best practice to balance the two metrics when defining personas. Source bio topic frequency helps find personas ultimately capture the topics’ breadth, while clusters show more granular topics overlaps and distinctions.
One of the most frequently mentioned source bio keywords for Crocs is “mom” which helped us determine a persona based around moms and kids:
TOPIC TREND CHANGE SENTIMENT POSTS APPEARANCES
TIP: while usually source bio data should take precedence in the audience research process, there are exceptions. If the original search relies heavily on source bio keywords, proceed with caution while analyzing source bio data.
This type of search will bias source bio results returned. For instance, if our search included consumers with the word “mom” in the bio our source bio, results would obviously return mom-related keywords like “kids”. Reporting on related keywords is a biased result created by what we searched.
Step 4: Consider how consumer language differs
Because of the way people talk online,, the next recommended metrics to review are top topics and topic clusters. These data points reveal the leading topics associated with your search and how they are linked to other key terms.
Here balancing frequency and clusters like source bios is recommended. When reviewing topics, look for unusual words not commonly used by the general population.
Also consider that relatively common words might be used more by a specific audience to drive a higher frequency for that topic.
For instance, “work” is mentioned way more frequently in our Crocs search results than we would normally see for the general population. This was a big clue that work or work-life was an important topic for one of our audiences.
Analyze words by groups instead of individually. For instance, in the Crocs cluster it’s hard to have a context for “sitting” on it’s own. However, when grouped with “sweats” and “casual” we start to hone in on behaviors associated with Crocs pertaining to a specific audience. The true power is in the action verb and surrounding context.
On a similar note, highly distributed positive or negative sentiment can be used to give more story and context to keywords. So, it’s important to consider the frequency of topics mentioned in tangent with its surrounding sentiment.
Finally, review hashtag frequencies and clusters. Similar to source bio’s hashtags, they are a way for consumers to self classify their own posts. With hashtags consumers are telling you how they categorize themselves and what’s most important in their specific posts.
In this hashtag cluster example we see how mom and kid hashtags cross-over. Kids may wear Crocs first, but moms may follow suit. This is key behavioral motivation for our “Crocs Moms” segment.
TIP: Proceed with caution with emoji data. While emojis are fun and our teammates at Infegy love to use them to spice up a report, studies have proven the cultural and generational context around emojis varies greatly. It’s easy to misinterpret emojis when they are treated as equivalents to specific words. The eggplant emoji might mean egg plant to your mom, but something else entirely in other contexts.
Step 5: Find demographic peaks
Our next recommendation is to look for abnormalities (think peak or troughs) in demographic distributions. These could be in gender, age, income, geography, or other demographic measures. It’s important to answer why these abnormalities occur. A unique audience might be responsible for a demographic skew.
Wearing Crocs skews higher for 25-34 year olds than general online talk about shoes. This group is more likely to be parents and adds to our case for a mom audience breakout:
Once audiences are confirmed, reviewing that their demographic metrics are appropriately skewed is a good way to double-check that your audiences are on target.
For instance, our Crocs mom’s specific audience search that we developed from our general Crocs query should have a 100% female distribution (no gender filter necessary).
Step 6: Understand potential social channel breakouts
Next look at channel distribution. Unusually high distributions on certain channels may be from a specific audience favoring a social channel.
Channel preference can be an important point to define for your personas. Knowing where they talk and interact will empower you to make actionable recommendations on where to target each persona.
Step 7: Add the finishing human element touches
Finally, as you have all of the insights gathered, you can organize the audience patterns into groupings. From there you can build out your personas with some qualitative research and post examples.
From the demographic research above, we landed on one of five Crocs brand personas, “Mommy and Me”.
Want to see how social listening with Infegy Atlas can help you build your own brand personas? Click here to get a customized demo with our team!
- March 2023 (2)
- February 2023 (4)
- January 2023 (2)
- December 2022 (3)
- November 2022 (5)
- October 2022 (3)
- September 2022 (3)
- August 2022 (2)
- July 2022 (1)
- June 2022 (1)
- April 2022 (1)
- March 2022 (1)
- January 2022 (1)
- December 2021 (1)
- November 2021 (2)
- October 2021 (1)
- June 2021 (1)
- May 2021 (1)
- April 2021 (1)
- March 2021 (1)
- February 2021 (1)
- January 2021 (2)
- November 2020 (1)
- October 2020 (2)
- September 2020 (1)
- August 2020 (2)
- July 2020 (2)
- June 2020 (2)
- May 2020 (1)
- April 2020 (1)
- March 2020 (3)
- February 2020 (2)
- January 2020 (2)
- December 2019 (2)
- November 2019 (1)
- October 2019 (1)
- September 2019 (2)
- August 2019 (2)
- July 2019 (2)
- June 2019 (2)
- May 2019 (2)
- April 2019 (1)
- March 2019 (2)
- February 2019 (2)
- January 2019 (1)