The advent of sophisticated social listening tools; tools that can crawl and analyse every tweet or every Instagram post ever publicly published, has opened up a wealth of opportunity for PR stories.
Take for example veganism, one of the hottest lifestyle trends that shows no sign of slowing down. A survey of 1,000 vegans would, firstly, cost a large amount (despite veganism feeling ubiquitous, it turns out vegan survey respondents are actually few and far between). Secondly, it would take ages to turn the survey around, and thirdly, as with all surveys, you only get the answers people want to put down, rather than knowing the full, unfiltered truth. Whereas with a social data report, you can analyse the 65million #vegan posts on Instagram to get a much more truthful and reliable set of data to analyse, within a much quicker turnaround time too!
Working with social data is brilliant; social posts offer stronger insights and contain information and opinions that people have freely volunteered. There are loads of social listening tools out there, but our preferred one is by our fellow Brightonians, Brandwatch. There are so many different ways to cut their data that you can always find something newsworthy to write about. But new types of data take time to understand and good social listening is a skill to be mastered, so I wanted to share some of the easy pitfalls that can be made when building a story from social data:
Thinking a hashtag only has one meaning
If you’re looking at social data to create a ‘Most hashtagged XXX’ story, be careful that your numbers aren’t skewed because that hashtag has more than one meaning. For example, if you’re looking at analysing what the most popular dog breed on Instagram is, you need to be aware that #yorkshire could be a post of a cute Yorkshire Terrier, but it could also contain a photo of the Yorkshire Dales or even a battered accompaniment to a roast dinner. When creating looking at social data, you have to be aware of the different meanings of hashtags and the context they are used in and make sure you exclude mentions that don’t relate to your topic.
Using words to judge sentiment
If you’re working on a story whereby you want to find out the nation’s sentiment towards something e.g. a TV show, a football player or a movement like veganism, then how people talk about that thing on social media is a brilliant way to dig out your stats. However, whilst social data analysis is great, it’s not perfect – especially when coping with the nuances of the English language.
For example, one tweet could say: ‘Neymar’s theatrics on the pitch are something else. Ridiculous’ and another could say, ‘Neymar’s skill at finding the net is something else. Ridiculous’. The first example is negative, criticising the footballer for his behaviour, whereas the second tweet praises him for his skill. Both use the word ridiculous, but the context changes it from an insult to a compliment.
Judging sentiment just on word-use isn’t a straight forward automated process. But there’s an alternative. Rather than relying on words for sentiment, when using social data reports we can instead analyse the use of emojis. We use emojis to help communicate the tone of our online statements, and to signal whether something’s said sarcastically or not. Emojis provide a much more accurate feel for how the nation feel about something … see our previous blog posts on the nation’s sentiment towards the different Love Islanders).
Location, location, location
Another thing to watch out for when researching a ‘Most hashtagged XXX’ story is the location of the post. For example, your client sells beach holidays and you want to investigate what the most popular beach activities according to Instagram are. Looking at the volume of mentions for things like #volleyball #kayaking or even #sunbathing won’t provide accurate figures for beach holiday activities, as these things can take place in plenty of other locations besides the beach. You would need to ensure your data only looks at posts which include a mention of the activity within a beach holiday context; it might take longer to set up the report, but the data will be more accurate.
Watch out for spam accounts
In a recent report, I was very excited to see the croissant, loaf of bread and baguette emoji all appear as the most used emojis around the topic of ‘Dubai’. It didn’t make sense and I was convinced I had stumbled across a hidden gem; maybe Dubai had an amazing bakery and patisserie scene that we could expose! Unfortunately not. The emojis had been used in an airline’s post to advertise a breakfast offer on their Dubai flight, and the post had been published hundreds of times. As with any data, you have to double check and interrogate your findings fully (don’t wait for the journalist to find the holes!). Consider looking only at original tweets and stripping out RTs for a truer view; or finds ways to strip out posts from organisations and businesses, to get a true consumer voice.
We’re constantly refining and tweaking the way we use our social listening tools, learning how to get better insights and stories from the each time. Check out some of our favourite stories we’ve created using social data…
…and feel free to drop us a line if you want to learn more about using social data for PR!