Big data is penetrable with the right understanding says Peter Sieyes, associate VP, head of consumer marketing and innovation at Infosys. Here he provides tips for creating a big data strategy…
Big data promises exciting new horizons for marketers, but few have managed to achieve them yet. It’s not because the data isn’t available or because the technology to collect and analyse it isn’t there, so what is holding marketers back, and how might they approach it?
Understanding the case for it
Getting the right message to the right customer, at the right time and in the right context are all increasingly easy, in theory, as customers’ digitised footprints provide data on a massive scale.
Search data, location data, online behavioural data, sentiment, social graph data and many other sources add potentially valuable information to demographic and transaction data.
The extreme volume, variety and velocity of all this data combined, is referred to as big data.
Obama’s last electoral campaign win, continues to be one of the most iconic cases for the power of big data capability in influencing human preference and behaviour. Campaign manager, Jim
Messina, built a talented analytics department five times the size of the previous 2008 campaign.
They combined data collected from fundraisers, field workers and consumer databases as well as social media and mobile contacts with the main Democratic voter files. They used cookies to track supporters online. Their massive data effort helped Obama raise $1 billion, optimised TV ad targeting and increased the effectiveness of door-to-door campaigning, phone calls, direct mail and social media.
They had a hypothesis their insights would focus their efforts on the voters whose opinion would be easiest to change, thereby optimising the effectiveness of their finite resources. They succeeded.
The challenges for marketers
No amount of data will get you anywhere unless you have the advanced analytical skills to manage and make sense of it. This challenge is true of ‘small data’, it just gets harder the more variety, volume and velocity of data you combine.
You need a hypothesis or a clearly articulated question you want big data to help you validate or answer. Without that you risk disillusionment. Both marketer and analyst need to work hand-in-hand from the very beginning of any insight development programme for it to have any chance of finding gold.
Many marketers are still stuck finding ways of correlating digital marketing investments to higher-order brand performance metrics. Leaders in this space are starting to develop new frameworks linking search data and sentiment analysis to existing brand metrics in order to enhance their understanding of what impacts current and future financial performance of the brand.
In the meantime many marketers are lower down the value chain looking at web analytics (traffic, dwell time, click-through rates), application interaction analytics for mobile, or likes, tweets and re-tweets for social. Four problems arise:
1. Most of these are efficiency metrics and not effectiveness metrics.
2. There is seldom a good understanding of how one area impacts the other.
3. The focus is customer behaviour within channel and not across channels, so there is no holistic view of customer behaviour.
4. The data is generally analysed offline, which means it struggles to give actionable insight that is timely. It is generally too late to leverage it.
However, marketers should not ignore the inroads these channels are making and should continue to push their learning from small data before assuming big data will fill the gaps.
The approach
To approach big data strategically marketers should:
• Shift their organisation’s attitude to data, from ‘justifying past decisions’ (codifying proven growth drivers), to ‘seeking new insights for predictive modelling or enhanced operations’. This can be a real challenge but it is essential if the implications of a major initiative require alignment to invest in new capabilities or to transform current operating models. A likely prerequisite may be the need for new systems and organisational change to remove data silos and to facilitate enterprise wide data sharing. These are major transformations.
• Define the higher order hypothesis they want big data to help them with and ensure there is a true link with core business objectives. It might relate to how different approaches to bought, earned, and owned impact brand performance.
If you are an ecommerce company it might be how the combination of structured, cross-channel transaction data, with unstructured online data, can optimise your next best offer to
your customers.
• Let no big data programme start without marketers and analysts working closely together from the outset to define the commercially-focused hypothesis.
• Ensure they have the right amount and depth of analytical skills in their organisation. If Obama’s 2008 data-heavy strategy required five times as many smart analysts in 2012, what is your starting point?
• Shift from offline analysis to real-time analysis. The market’s best systems solutions have this capability.
• Optimise the data you already have and don’t think it always has to start with
big data.