Predictive Behaviour in Social Media

1st June, 2015

One of the central tenets of market research is to record behaviour, and make predictions based upon analysis of those records. At the core of this is data. Social media has altered our relationship with data; there is now more of it than ever.  

So how does this colossal rise in information affect market research?

Social media channels now comprise over 2 billion active accounts, of an active internet population of just over 3 billion.

This means roughly 66% of all internet users are engaged with some form of social media. The transition in data is phenomenal. Every two days we generate more data than the entirety of the world did in the whole 2002 – something like 5 Exabytes* – and this figure is rising.

Predictive Social Media – the end of the humble survey?

Social media mining, ‘listening’ or ‘scraping’, grants market researchers access to a vast opinion pool that is constantly recharging and churning out more information.

Social media mining can be extremely useful and you likely already engage in this practice, perhaps without realising. Social media management platforms such as HootSuite enable us to monitor our favourite trends, hashtags, pages and more in a unified web app. Social media mining and listening work similarly to this.   If we just take Twitter – access to a sample of 302m users is phenomenal – and if we factor in the other 1.7bn social media accounts, well, you’ve got yourselves a seriously large opinion pool stemming from four corners of the globe.

Its increasing use has led prominent market researchers to state that in the coming years the survey – a most trusted tool – will eventually die out. The theory goes that the survey will slowly be replaced by efficient and accurate social media listening platforms that can make market predictions based upon qualitative and quantitative data, extrapolating core information via filtering and algorithms.

We have to be realistic, however, about the supposed prowess of utilising social media mining to garner truly representative measurements. Social media platforms can easily be gamed by those wishing to alter common perceptions, whilst other events might cause an idea, or theory, or product to “go viral,” but leave us without a real indication of why this happened.  Also there will always be the gap in reality between what people say they’d do versus how they behave in practice.

The Internet of Things

Research data will be further embellished by the vast number of devices coming online. The coming years will see the Internet of Things provide active feedback to market researchers. Predictive engines will gauge and weight data from a combination of sources to more accurately predict upcoming trends in the wider world.

This isn’t limited to the wider social media network. The Internet of Things, along with social media mining and listening, can be refined to a smaller area to extract meaningful information from local chatter and behaviour.

Using social media to predict market trends will still rely on conversations for insight – we just have to approach the conversation differently…


* For the geeks among us, an exabyte is a multiple of the unit byte for digital information. The prefix exa indicates multiplication by the sixth power of 1000 (1018) – or a billion gigabytes – that’s a lot of data…