We featured in the Business Reporter report on Retail (distributed by The Sunday Telegraph) yesterday. The article discussed the power of social media to predict future demand in e-commerce through predictive analytics of data mined from social media.

Below is the full article.

Can social media predict the future?

What record jumped 262% from 6,939 to 25,088 in 2011?

What first took 3 years, 2 weeks and 1 day now takes less than five days?

The answer to the first is TPS - that's Tweets per Second. The first record was set on 1st January 2011, 4 seconds after midnight, in Japan. The second, also in Japan, in December the same year when Hayao Miyazaki's cult movie "Castle in the Sky" was shown on TV.

The answer to the second is the time taken to send one billion tweets.

Social media, in particular Facebook and Twitter, are the most active conversations on the planet. Where Facebook is younger and gossipy, Twitter, with its older demographic and 140 character limit, tends to be more direct and factual. This has made the micromessaging channel the main focus of researchers looking for ways to deepen customer insight to improve their product offering and predict future demand.

This is proving particularly interesting in politics where companies such as Mashable and Globalpoint are examining social media messages, asigning semantic meaning, and trying to predict results of the Republican primaries. They are having some success, in particular in spotting swings, but it's early days and attaching meaning to informal social language is difficult.

Other companies are looking to how this can be applied to eCommerce. Colbenson, makers of search engine analytics tools, are used to working with user language to try and connect potential customers to the products they search for.

"It's hard, customers think differently than sellers," says Ángel Maldonado, founder and head of Colbenson, "we always start by looking at how users search and how much gets little or no success"

"Social media can give us insight into their behaviour, we can have search terms already hooked up to results with proven track records for conversion"

But how about predicting the future?

The theory runs that what is talked about today on social media channels gets translated into demand tomorrow.

"Semantics is less important in eCommerce, nearly 90% of user product reviews are positive, so it's a reasonable assumption that notoriety of catalogue items on social media will translate to demand"

According to social commerce experts Bazaarvoice, some 53% of Twitter users talk about products and offer recommendations, and 48% of those deliver on purchases.

This makes Twitter a powerful reservoir of information about potential purchase behaviour and demand.

"The thing is," Maldonado says, "No one knows what they're missing out on, business don't measure the empty spaces: who came in the store and asked for something that's sold out, or who searched your site but couldn't find the product. Business measure sales - supply. Twitter measures demand"

Ecommerce is particularly well poised to take advantage of this opportunity - with the UK being the largest eCommerce market in Europe.

"This is not just about predicting an increase in umbrella demand when see the rain, it's about a demand you didn't anticipate or a connection you didn't make", Maldonado explains, "At Colbenson we've been looking at how to respond when a catalogue item is mentioned next to something we didn't think of, maybe Julia Roberts was seen in a pair of Hunter boots and those four words show up time and again "Julia / Roberts / ‘Hunter boots'" - it won't be long before that search hits your site and you need to connect it to the right product or you've lost a sale or maybe even a customer"

This research is still in its infancy, but there are promising signs that data mined from social media can add real value to online business by improving customer insight and predicting future demand.

Here is the link to the actual article in the Business Reporter Retail report.