Experienced data researchers know this process always starts with a good question, but not every brand is able to do this at first. There is a way to conduct a trend analysis using social media data where you start with a general hypothesis and find a much better one half way through the research process. The point is more that the process gets started rather than waiting for genius to strike.
Setting parameters when conducting a trend search feels counter-intuitive, after all, isn’t the point to look at everything to find a trend? But by drawing a line around an area, such as a physical location, you’re controlling for a better understanding of the results. Depending on the question at hand, social media data also allows for parsing by gender and age, though most researchers will keep that net open to start.
Scoping the analysis by topic is the most crucial part, since this is where conclusions will be drawn from the qualitative portion of the data. It’s tricky because topics are created using information that is known in search of information that is not known, so be prepared to do this process several times.
Data sources are also extremely important to think through. Pulling results from one social media network at a time tells part, but not all, of the story. It’s much better to use a multi-source solution when analyzing social media data for broad trends since a majority of social media users are active on two or more sites.
Once these steps are in place, the information coming in can lead in several different directions. A few common insights that can surface during this portion of the research include:
- Unknown acronyms: for those living and breathing the trend every day, it’s common to want to shorten frequently used names, phrases or references. It’s also a way of taking ownership over something. For example, people who frequent, and live, in Outerbanks beaches in North Carolina, affectionately refer to the area as “OBX.”
- Sentiment associations: while the original query probably did not contain sentiment related keywords, if there are any strong feelings around the topic, they can be uncovered and quantified during this stage, though it’s best to hold off on a full sentiment analysis until later.
- Unique product usage: sometimes brands are surprised to see images of people using their products in unintended ways. Makers of the Boon drying rack found images of their plastic green square being used to dry wine glasses instead of baby bottles.
After making a note of the quantitative data in your original search, go back and integrate the new findings into the search queries – or considering revising the overall trend search. Also make a note of any interesting findings that might affect the research’s scope or demographic parameters. If an age group is skewing results away from getting concentrated findings, it might be helpful to exclude them.
It’s also a good idea at this point to step outside social media for a moment and gain a supplemental perspective by looking into Google searches, Amazon reviews, etc. to acquire a multi-dimensional way of the looking at the newfound topics. There might be a keyword or sentiment worth adding back into the search.
Once a certain level of familiarity is gained, pull the most popular content by various types in different combinations. A couple of suggestions include:
- Image content only
- Text content only
- Text with links only
- Text with links posted by men
- Images posted by women under 35
Before reaching any conclusions, take a few moments to examine the profiles of the influencers in this space – these could include private individuals or public organizations. Ask how their content contributes to the story being told about this trending topic.
Finally, always remember to take snapshots of the data along the way. Sometimes the sheer volume of content is the story by itself. Conducting a trend analysis utilizing social media data, including looking at brand-specific information, is a great way to know the context of conversations and reasons for rising trends.