‘The Facebook’, as it was originally known, first appeared in 2004 as a way of capturing online profiles of Harvard undergraduates. It came into mainstream use in 2006 and since then brands have struggled to tap into the real value of social data and use it to make informed business decisions. Marketers analysing other social networks have encountered the same problem.
Many of us will still remember the days of counting mentions, likes, hashtags and re-tweets to establish the success of a campaign. You might think that those days have long gone, but sadly you’d be wrong. I’ve seen a number of recent campaigns where ‘success’ has been gauged by this rather crude metric.
According to CMO research carried out at the beginning of 2014, 70 per cent of CMOs can’t quantify the impact of their digital campaigns and 70 per cent of marketing projects don’t make use of the available data. Yet there’s no real reason for this to be the case. There have been a number of technology innovations over the past few years that enable a shift from seeing social media as quantitative data to qualitative data at scale.
Social media analytics – the early years
Much of the value of social data comes from what people are sharing and how they express things. At DataSift, we call it Human Data, covering the whole spectrum of human-generated information – whether text, image, audio or video – shared with others via social networks, blogs, comments or news sites. Analysing such information effectively is no mean feat and it’s fair to say that early social media analytics didn’t come close to doing so.
It began with counting Facebook likes, or retweets on Twitter. But this can’t really provide true insight – it’s no more than a vague online show of hands. What brands wanted to know (and still want to know now today) was what people were thinking and feeling, what’s working and what’s not working when it comes to campaigns, and how best to engage with their audience. With correct analysis (carried out in a privacy-first way), Human Data can prove valuable for market research, gauging brand health, marketing optimisation, measuring customer satisfaction and more.
So social media analytics got progressively more sophisticated. Counting likes evolved into content analytics software, with sentiment monitoring that allowed brands to begin making sense of the unstructured information that constitutes Human Data.
What makes good social media analytics?
With the technology starting to develop in line with the requirement for meaningful social media analysis, what makes good social media analytics in 2015? What distinguishes what can be done today, from what passed for analytics just a few years ago? DataSift isn’t a social analytics company, but our technology powers many organisations that are and I believe that the differences today can be distilled into these three main elements:
- Depth and sophistication: it’s really all about the depth with social data analysis. Sentiment analysis is a powerful concept, but social media analytics takes into consideration factors such as context and content in addition to sentiment, to provide a depth and sophistication of insight unrivalled by any analytics that came before.
- Real-time: an hour is a long time in social media, so analysis of Human Data has to be real-time to be effective. What is relevant in the morning might not be later that day, so it’s imperative to analyse data immediately.
- All content holds value: although Twitter and Facebook are the two poster-children for social data, there is also immense value in other social networks, and also the comments that people leave on blogs and news articles. At DataSift, we work with more than 20 social networks and the more content analysed (including audio, video and images), the more powerful the analysis will be.
Thankfully social media analytics has come a long way, even though many brands are still stuck in the qualitative dark ages of counting likes and mentions. Come into the light guys, quantitative analysis will give you more than you ever thought possible.