The use of social media data analysis has expanded exponentially over the last ten years, and for a good reason. Social media platforms like TikTok, Instagram, and Snapchat have become an integral part of our culture, boasting billions of active social media users globally.
Today, consumers have no problem adding personal information to their social profiles and publishing content about their personal lives, interests, what they love, and what they hate. This data is stored on the internet and given to advertisers to target with campaigns. Social media data is also becoming easier to work with as new technology is helping researchers manage and analyze it more efficiently.
Before diving into various use cases of social media data, let’s first talk about data formats.
Understanding Different Data Formats
Data can be separated into three main categories: structured, unstructured, and semi-structured.
Structured data is stored in a database that organizes information with strict formats for each piece of content. This type of data is most accessible for future trends or behavioral indicators. Examples of structured data include names, dates, addresses, locations, credit card IDs, stock information, etc.
Over 80% of data is unstructured. Unstructured data is typically text, photos, or videos not organized in a database. This type of data can be beneficial, but it requires automation tools to structure, process, and analyze. It is also more challenging to mine the trends because there aren’t defined fields.
Semistructured data is stored in a database, but it must be parsed according to some criteria set forth by an analyst. For example, RSS feeds are kept the same way as unstructured data because they do not have defined fields. This data type usually has a hierarchical structure, allowing for simple analysis to find trends or variables within a subset. For example, a social media directory would include weakly semistructured data because a particular article may only include the name, URL, and a short description. The hierarchical structure gives marketers more clarity on the social media data used in the network analysis.
What is Social Media Data?
Social media data is information users publish on their personal social media channels. This includes name, age, gender, language, location, job titles, and how they describe themselves in their social media bios. It also has who they follow and the daily posting of text, videos, and photos representing their interests and sharing them with their online business communities. Social media analytics is valuable to marketers looking for audience insights that may drive brand engagement, increase sales, leads, or conversions, build brand reputation, or change consumer behavior.
Over the last several years, I have documented a five-step process in using technology and social media analysis and creating a competitive advantage.
Starting with social media analysis design
Designing your research analysis should always be the first step in any project. This requires asking critical questions: What is your analysis trying to accomplish? Who will do the work, and how much time will they have? What data will you be using? What is your procedure for coding and reviewing the results of your analysis? What questions do you want answers to?
These are all questions that must be documented and answered before getting started. Many projects fail because this step was skipped or there wasn’t adequate time spent thinking through the problem you are trying to solve with social media analytics. It’s not uncommon for this process to take 3-4 weeks.
Social media data collection is critical
Data collection is the next step. This requires scraping social media data across all the platforms and structuring and organizing it. This is easier for some social networks than it is for others. For example, most social listening platforms have been scraping Twitter since 2007. They have full access to what Twitter calls their fire hose. Other platforms can scrape URL data from articles, forums, and review sites, which is also beneficial.
Unfortunately, Facebook, Instagram, and LinkedIn have very strict APIs, so scraping their data is difficult. Some scraping tools can do this, but I would be careful using those because they might be against each social network’s terms of service. In this step, the most important thing to consider is determining what questions you want to answer and identifying the correct data source.
Social media data mining is time-consuming but important
70% of the time spent on a content analytics project should be spent mining the data. This requires a human filter, manually searching, curating, categorizing, and reading through social media data. This also involves structuring the data to be easily uploaded to a data visualization tool or creating a pivot table.
For example, it will be much easier to parse through Twitter data because they have an API in your project. If you are scraping from forums or review sites, you may have to search for specific keywords and phrases within those sites. So you will need to identify the correct search terms for your hypothesis and build a complex Boolean query to capture all the social media data.
The result of social data mining is that the analyst can more easily contextualize the insights based on what they see in the data. The beautiful thing about this is that you don’t have to be a data scientist or even an analyst to mine social data. There are tools available to help.
Social media analysis provides context
Some might say that data analysis is the same as data mining, but it isn’t in this context. Data analysis tests a specific hypothesis and translates those findings into insights.
For example, let’s assume that while you are data mining, you notice some conversation about a specific product. Your hypothesis can go one of two ways. Perhaps it was an anomaly, and the audience only mentioned it once, and you found it while mining the data. Or the audience talks about it a lot. This type of data can uncover actionable insights.
To test the hypothesis, you might create a filter and look through the data for all product mentions to see whether conversation volume is high or low. If the volume is high, your insight is that this audience talks a lot about the product, and you may even go one step further and start tracking brand sentiment.
Social media data visualization helps marketers tell stories
Many analysts and BI engineers work with data visualization tools such as Tableau, Qlik, Google Data Studio, or Datorama. These tools allow them to quickly see the insights within their social media data and test hypotheses based on those findings.
Tableau is one of the most commonly used data visualization tools. It allows for easy configuration and changes to be made with drag-and-drop functionality. Within Tableau, you can visualize different data types, including line charts, bar graphs, and bubble charts.
Google Data Studio is another popular data visualization tool. Users can build reports with Google Data Studio, including data charts and pivot tables. In these reports, you can also visualize your social media analytics to discover patterns or significant relationships between variables.
While Tableau and Google Data Studio are popular data visualization tools, many free online options can be just as powerful.
Social media data visualization tools are good for telling a story using charts, graphs, and other visualizations. It’s also beneficial because it allows others to view and interact with the data visualization. It’s important to note that data visualizations may also be referred to as a social media dashboard or scorecard.
Social media data reports & scorecards
There are two main social media data reports: Campaign Performance and Social Media Scorecards.
A social media scorecard is a report and visualization that uses metrics to determine an organization’s success in each of the three areas of social media marketing — relationship, listening, and engagement. Generally, these are assessed by audience size, share of voice or conversation, social media engagement, and customer care metrics like complaints received. Many social media scorecards are designed around monthly or quarterly reviews that assess progress toward pre-defined goals.
Social media campaign performance involves identifying, collecting, organizing, analyzing, and reporting on a program or campaign. It evaluates, measures, and interprets social media results, including financial metrics like sales and ROI. It will also have social media data like clicks, CTR, impressions, clicks to the website, conversion rates, and unqualified leads.
In any case, exporting the data into a social data report is the right thing to do. It depends on your audience’s sophistication or whether they have a license to Tableau or other data visualization tools. Whatever you choose, it’s always a best practice to visualize social analysis using social software.
The great thing about putting the data into a report is that you can control the whole narrative and tell a story from one slide or page to another. The bad thing about putting it into a report is that some people might not read it in the file sizes are too high.
The biggest takeaway is this: brands must use social media data to inform how they engage with audiences. This can go a variety of different ways. You could use this data to inform headlines, social media content, and blogs. You can also prioritize media relations strategy based on what the audience reads and shares online.
The marriage of social media analysis & survey data
Social media analysis studies social audiences, content, and interactions on social platforms to understand how people use them and what they’re talking about. On the other hand, survey data is a quantitative measure of a given population’s attitudes, opinions, or behaviors. It is gathered by administering surveys to a sample of that population.
Both are important because they provide insights that can be used to improve everything you do from a marketing, communications, product, and innovation perspective. They can be used to uncover market white space, drill down on audience behaviors, and gain insight into specific questions you want answers to.
Social media analysis is complicated mainly because it’s newer, and the innovation of social networks continues to evolve. I look at social analysis through two different lenses.
The first lens is performance and optimization. When brands use social media as a paid advertising channel, they must track performance, optimize spending, and adjust. This is paid social analysis.
Social analysis can also better understand an audience, content strategy, or conversation about a topic or brand.
Social Media Analysis Examples
I used to believe that a social analysis would tell me the “what” and that primary research would tell me the “why.” Here’s an example from the Prophet Brand Relevance Index report on the Top US Brands. They used primary research as the data source of the survey. I would expect that a combination of social data would change the results.
Think about it. I can understand what media publications a particular audience reads more than others with a solid social media analysis methodology. With primary research, I can ask why one media publication is preferred over another. What I have learned throughout my data journey is that you can get both the “what” and the “why” from primary research and social media analysis.
Let’s take the Brandwatch Customer Loyalty Report published in 2021 in partnership with Global Web Index as an example. Before we look at the shopper journey data, we must first define Brandwatch as a social media intelligence platform and GWI as a research company that uses primary research to generate consumer insights.
The report leverages insights both from social analysis as well as survey data.
Before looking at the data, let’s read about their approach first. It’s a perfect descriptor for using primary research and social media analysis combined to generate powerful results.
Social analysis and survey data are inherently complementary as the former offers an unfiltered view of unprompted consumer opinion over time on almost any topic enabling researchers to find unexpected insights. Primary research allows researchers to ask consumers the exact questions they need answers to.
So again, combining the two data sources allows researchers to ask questions about unexpected themes cropping up from the social analysis. When looking at the second paragraph here, GWI tracks the number of brand advocacy questions every quarter, and the question they ask is, “What would motivate you to promote your favorite brand online?” The top three answers are:
- High-quality products
- Great social customer service
Looking at the right, we’re looking at the social analysis. This required a significant Boolean query to qualify certain conversations into these buckets already predetermined from the GWI survey. In this case, they are using social media analysis to validate the findings from the primary research.