Big data has been one of the most hyped buzz words over the past year or so, appearing in almost every trend analysis in every single industry. As Wikipedia puts it can be used “to spot business trends, determine quality of research, prevent diseases, link legal citations, combat crime, and determine real-time roadway traffic conditions.”
A bit over the top? Yes. But despite the hype and the post-hype backlash, Big Data is here to stay. And it’s here to stay because it has value.
Big data and the Obama Campaign
Many words were written about how the Obama 2012 presidential campaign used big data in order to secure four more years in the White House. A team of 100 data analysts tucked away in a windowless “cave” and an obsession with “measuring everything” were the foundation. The true magic however, lies in how campaign manager Jim Messina used insights derived from the data to execute roll out countless campaigns and targeted activities with surgical precision. Every donor, volunteer or potential voter heard exactly what they needed to hear in order to mobilise them at every step along the campaign.
At the beginning, the Obama Campaign suffered from a problem that many CSPs suffer from – disparate datasets about their constituents, containing different types of information. The Obama campaign realised that broad segmentation such as age, gender and geographic location are insufficient for the type of precise messaging required. A holistic view of the voters was needed for a much more personal and effective campaigning approach.
The first step they took was to combine all data they had about their voters in different systems. Nothing was ignored. Everything from donation history to e-mail opening rates was included. This broad, view of voters enabled the use of Big Data methods to identify patterns and trends and was the instigator of some very specific and very effective micro campaigns.
Hidden treasures in the data
One such example is an A/B tests of Obama’s messages aimed at women. The data showed that the voters most responsive to the campaign’s arguments about equal-pay measures and women’s health were actually those whose likelihood of supporting the president was low. These findings suggested that here was one thing that could pull conservative woman voters over to Obama. A direct-mail campaign addressing women’s issues was immediately launched. It targeted conservative women who were at odds with their party on gender concerns.
A marketer’s dreamland
Brands everywhere are realising that segmentation by payment method, demographic, age or business type does not truly reflect the individual preferences or needs of customers. CSPs in particular can generate an extensive view of their customers and by using the same big data approach as the Obama campaign they can uncover the person behind the data and achieve great results. Seems like a marketer’s dream come true: creating continuous engagement, maximising brand image, customer satisfaction and revenue by giving customers exactly what they want. Is it really possible?
When dreams hit reality
In theory, big data makes marketing to a Segment of One possible. Reality is a bit more complex as turning the big data insights into action is yet another challenge.
Take for example Karen. She is in her 20s, has a smartphone and is a social media maven. She has a package containing 300 minutes and 750MB of data, which she usually uses up after three weeks. She doesn’t upgrade to a larger package because the next available plan is 600 minutes and 2GB of data and costs 30% more. Since she is stingy with her budget, every month she bites her lip for a week or so and suffers the slow connection. Anna, Karen’s closest friend with whom she speaks every day at least twice, just switched to another operator offering a deal better suited to her needs.
A proper big data analysis would flag Karen as being at a high churn risk. Her package does not fit her usage pattern; she called customer care to inquire about a larger plan but did not complete a purchase, and the number she dials the most has just churned.
A Segment of One treatment would mean having customer care contact Karen and offer her a plan with 300 minutes and 1.2GB of data, priced somewhere between her current plan and the next one up. Alternatively, the next time Karen maxes out her data, it would immediately trigger an SMS suggesting a week long data package of 400MB that should last her until the end of the month.
By now most CSP marketers are shaking their heads, knowing that this scenario is almost as fantastic as Hogwarts. Packages that are out of sync with the bill cycle and ad-hoc plans that are not predefined in the product catalogue don’t exist. Any effort to create such an offering would require a definition and implementation process that takes the real out of real-time. In most cases, by the time the offer is up and running, Karen would be long gone.
From analysis to fulfilment
Big data analysis is an important part of current day marketing, but it is only part of the equation. An equally important part is the ability to provide an individual subscriber a tailored offer in real time. An end-to-end solution that handles the process from data gathering through to fulfilment is a must, or big data will remain just that – data.