4 Strategies for Getting Social Data Analysis Done in Less Time

Get social data analysis done in less time with these proven expert strategies while also increasing the quality of your insights.

Analysing social data can be time-consuming but with experiences come strategies to cut down on analysis time that interestingly improve analysis quality, four experts share their best strategies for getting back more time.

Remove the Noise from Retweets on Twitter

David Barrowcliff, Research Director at Newton Insight, says that when analysing Twitter data you can save a lot of time and pain by simply removing the retweets.  

More often than not, RTs add a lot of unnecessary noise and tend to get in the way of the underlying good stuff. You’ll find that your data is less noisy, cleaner and not skewed to one or two ‘celebrity’ opinions.

Depending on the questioning you are answering, you might want to include retweets in your data but for most experts, they add too much noise and can skew analysis results.

Focusing on First Person Conversations

• Tamara Lucas, Associate Director at Convosphere says that focusing on first-person conversations gets you to the best insights faster by removing as much of the nose as possible.

Noise led by users sharing news (not necessarily by retweeting them) makes it hard to find “personas” conversations or opinions. One of the most time-saving tricks I use it is to include a filter in every project to gather first-person conversations by using the pronouns or common verbs forms that express opinions or are related to the topic of analysis in both singular and plural first-person.  For example, if I am looking for patients conversations, I normally add keywords around the verb “to suffer.

By using first-person filters Tamara can focus and spend more time analysing what people are discussing or commenting.  This reduces the time cleansing data and the time spent just reading about the same news over and over again.

Put Tamara’s strategy into action by researching how people express opinions around your topic (brand, product, service etc) and create a filter in your data set to gather those specific conversations.

Reduce Bias By Leaving Assumptions At The Door

For Adam Brons-Smith, Social Listening, Social Media Analytics Consultant at Brandwatch strategies for removing analysis bias can majorly cut down on analysis time.

He advocates for taking a step back and following the data to let social insights shine.

I think the biggest thing I’ve learned, has been to approach social data with no assumptions of what’s going to be there. One of the biggest mistakes is having preconceived ideas, as when it comes to analysing the data you can allow yourself to have a bias.

This is a great tip when you’re project is to ‘validate’ an idea or approach with social data.  Reframe the question in your head so you’re analysis is not biased to validating – you should be testing.  By testing, you can follow the data and maybe even find an insight that would be otherwise hidden.

And finally, from me…

Create a Lexicon Bank

Me.  I’m Dr Jillian Ney, founder of The Social Intelligence Lab.  

Over the years, I have created lexicon banks on topics. These banks are simply lists of related words or phrases that are commonly used when people discuss topics online.

For example, if we’re looking at the weather you might has the general sun, rain but expressions like clammy or freezing are also important.  The worst thing is stilling with a blinking cursor trying to write a query or segmentation filter.  If you’re not prepared, there is the tendency to just put in words but they are not necessarily the words that people use when talking online.

I second Tamara’s approach for the first-person conversations too.  Many of my lexicon banks get more complex and contain first-person filters and phrases.  It’s also useful to have banks for different countries, territories or even cities.  Thinking about something simple like “pants”.  The term “pants” mean different things in different places.  Also, remember to account for slang and colloquialisms.