Big data, like many IT terms, has fallen victim to the hype bug. Ask any vendor what it is and you will likely get a different answer. You'd be forgiven for thinking big data was all about the data size, but it's rather about the volume of data, the velocity of the data, and the compute power required to handle such and make sense of it. So, for example, sensors in a field of wheat recording temperatures, soil composition, wheat health and so on will send lots of small bits of data, often, to whoever is doing the monitoring. Making sense of that in real-time requires a whole lot of crunching power, which computing systems haven't been able to do - until now.
Says Jean-Marc Bonnet, head of applied innovations and intelligence at Teradata: "Today's big data will be small data in five years' time. The real issue is what you do with this data. Conceptually, when we talk about big data, we're really talking about what insights we can get from it, so we're really talking about data analytics. New types of analytics are enabling us to process these huge volumes of data to come up with profound insights. They're enabling us to build on the statistical analysis of transactions to interpret interactions. That's easier said than done, because this quantity and variety of data creates a lot of noise; the trick is being able to identify what's relevant for the analysis."
The right questions
KPMG IT advisory director Frank Rizzo agrees: "Having the right data is more important than having the most data. Once the right data has been identified, analytics can then be applied. Businesses today are under the impression that analytics is all about the software and tools used. Rather, it should be seen as a process whereby better decisions can be made by asking better questions. "Without analysing the right data, however, this process becomes difficult," he says. "The value of big data to an organisation is measured by the insights that management can draw from the data. To get to the insight, we need to be asking the right questions and ensuring that the data has integrity. Data and analytics isn't the newest business trend ? it's a permanent change in the business landscape, one that must be understood."
SAP head of database and technology and mobility Manoj Bhoola says companies locally and globally have caught on to the concept of big data and the fact that they need to do something about it, but many aren't sure, as yet, exactly what to do with this data.
A few obvious applications come to mind, like tracking sentiment via social media, where big data analytics are being used to great effect. Others are still dipping their toes in.
Relevance
Says UCS Solutions senior executive, business solutions, Jessica Knight: "There's a jumping-off point for retailers with loyalty programmes - they have a means to identify customers so they know who they are, where they shop and what goes into their baskets."
This, she says, means retailers can use the data they have been collecting via loyalty programmes for all these years and target promotions, for example, more effectively. Says Knight: "Clicks Clubcard, for example, has a lot of historical data it hadn't done much with so far, until two years ago, when it decided to be more thoughtful in its offers to customers to get them to return to the store, or spend unused points."
Clicks crunches its historical data to find out, for instance, which customers had previously bought perfume/aftershave and then targeted them for a Father's Day promotion, which got a good response because it was relevant.
"Clicks did 32 promotions as a pilot and managed to improve the average response per mailer from an average of 5% to an average of 13%, with one mailer getting a 54% response," Knight states.
Endless possibilities
The power of big data, says Raj Wanniapa, GM of big data for Dimension Data EMEA, is in combining different data sources to gain insight.
For example, if you combine data on people movements with health data, you can more effectively manage the spread of disease by working out where they came from and who is likely to be carrying them and where to.
Knight says some organisations are starting to leverage third-party data to make more sense of the data they own and to gain deeper insights, for instance using credit information from the credit card issuers with loyalty programme information gathered by a retailer to track spending patterns.
"We've worked with retailers using a combination of mobile, credit and retail data to look at millions of customer transactions over two years and establish which customers are the most valuable and which the most loyal," Knight says.
New types of analytics are enabling us to process these huge volumes of data to come up with profound insights.
Jean-Marc Bonnet, head of applied innovations and intelligence, Teradata
The future, says Patrick Evans, director of newly formed Fraxses Africa (Fraxses is a BI metadata platform that does big data analytics), is in platforms that can analyse millions or billions of records from disparate sources in nanoseconds, and present the data to business users in a format that's people-friendly. From fraud to perishable goods management to healthcare and agriculture, the possible uses for big data are endless.
Big data in action at the polls
Digital marketing and media company Apurimac Media researched which political parties made use of social media platforms such as Twitter and Facebook to interact with their supporters in the lead-up to this year's national and provincial elections. Its research started at the beginning of March, and weekly surveys concluded after voters went to the polls on 7 May.
"We conducted this research to see whether or not political parties effectively made use of these channels, and if it translated into any improvement in results," says Apurimac CEO Will Green. The desktop research was conducted independently and was not paid for by any political party. South Africa has around five million Twitter users and nine million Facebook users, against a national population of just over 50 million.
The political parties that best integrated social media into their existing communications plans saw the best results, says Green.
"The Democratic Alliance, with its #Ayisafani TV and YouTube campaign was a big hit, and the initial barring by the public broadcaster of its flighting worked in the party's favour," Green says.
"And the ANC cleverly spotted the obvious synergy between political rallies and digital media, when they made WiFi available inside FNB Stadium for the Siyanqoba rally," he comments.
Having the right data is more important than having the most data.
Frank Rizzo, IT advisory director, KPMG
Green says the lack of available or allocated human resources and experience failed smaller parties, something they can learn from for future elections.
Green says Aprurimac's research found that whereas political parties performed better on Facebook platforms, it was the individual political party leaders and party officials who performed better on Twitter. Here, he says, a new form of 'baby kissing' emerged, where party officials took time out to respond directly to tweets.
The ANC recorded the biggest Facebook growth, over the research period, growing from 52 536 likes to 136 046 from 7 March to 2 May 2014. The ANC's Facebook audience overtook its Twitter audience and eventually also overtook the DA's Twitter audience, which was the biggest of all parties at the beginning of the campaign.
When looking at the percentage growth in Twitter followers for political party figures, Mmusi Maimane topped the list, with 15.5% (from 27 800 up to 32 900 followers). Tim Harris grew his audience, with 10.18% - up from 9 935 to 10 800, followed by Lindiwe Mazibuko, with a growth of 6.8% (from 120 800 to 129 700), Floyd Shivambu's 6.2% (from 42 400 to 45 200), Patricia de Lille with a growth of 5.8% (from 35 600 to 37 800), and Zwelinzima Vavi's growth of 5.5% (from 157 800 to 167 000).
Share