3 clues that Big Data might be Old News #SmallDataForum

I had the very great pleasure of co-hosting a roundtable on data and insight for LexisNexis yesterday with one of pioneers of podcasting and all things digital, Neville Hobson, @jangles, now an integral part of IBM’s social consulting team. Refreshingly in a world where we’re routinely made to feel insignificant and overwhelmed by the very bigness of Big Data, the session was titled (and hashtagged) as the #SmallDataForum.

CLUE 1: All of the 25-strong audience have roles that are shaped by the need to manage and extract intelligence from enterprise scale data sets, on behalf of their own organizations and for their clients. And it was encouraging to see and hear that none of them are remotely intimidated by the task at hand and the skills they need of themselves and their teams. How different the feeling in the room from other, similar events I recall from just three or five years ago.

Back in that day, talk of exabytes (1018) and zetabytes (1021) of data bought CCOs and CMOs – let alone CIOs and CTOs – out in a cold sweat, even if they had some idea of how many zeroes these big numbers entailed. They’d have blanched and quailed at my story from the critically-acclaimed book Big Data, A Revolution That Will Transform How We Live, Work, And Think. Authors Viktor Mayer-Schönberger and Kenneth Cukier calculated that, by the end of 2013, there were an estimated 1,200 exabytes of data stored on earth. The sum of human knowledge, production and data; enough to fill 40bn, 32GB iPads, which would stretch to the moon. All this data and so few insights. And we produced the same volume of data again in 2014 alone.

CLUE 2: Yesterday’s discussion – hosted and smartly framed by LexisNexis’ Thomas Stoeckle, the company’s Global Head of Evaluation and Insights – was very much more about how we can best identify, isolate and bring together all different sorts of data, and then make sense of them. The consensus was, even though this can often mean tens or hundreds of thousands of lines in a spreadsheet or a database, when we get our hands on the data we need to understand our businesses and our markets, we’re only often only dealing with a small backwater – just a puddle – from the vast oceans of potential Big Data. Potential, yes, but without smart search and sift and analytics, potentially a meaningless morass of more noise than signal. The concept of little big data, or small data, really seems to be catching on.

CLUE 3: There was also broad agreement that data per se, however artfully automated, is fundamentally just ones and zeroes without proper human intervention and analysis. While we weren’t of one mind where the key intervention points from human analysts lie on the journey from data to insight, all agreed that humans and machine analysis are yin and yang. Plus we had an opportunity to share and laugh with Carrie Fisher at her – and IBM Watson’s – “therapy session for robots”.

And perhaps the most revealing answer to the group exam question “What single piece of data processing will make most difference to your organisation this year?” was “Cleaning up our salesforce.com database”; a real bit of small data activity deemed most likely to yield the most significant business results. Quite so.

Our discussion was also rich in metaphor. Stimulated by Vlifredo “Mr 80/20 Rule” Pareto’s other important maxim – “An idea is nothing more or less than a combination of old elements” – a delegate compared successful small data analytics to the process of baking a cake. You can buy all the very best ingredients, but if you don’t know how to put them together in the right way and the right order and cook them for the right amount of time, you’re always going to be closer to garbage than gâteaux.

And in the debate on whether data wrangling is more oil than soil, I couldn’t resist citing my mentor, Pierre Emmanuel Maire, the Ammirati Puris Lintas planner who developed Unilever laundry’s compelling “Dirt Is Good” insight, that using data to look for insights it like prospecting for oil. Over lunch last year, he told me: “Unearthing genuine insights is like finding oil. First, you zone in the right area. Next, you mine in the right place. Then, you extract something relatively crude. And finally, you refine it until you have something powerful.”

The nods around the room at that suggested that, while we’ve clearly got a long way to go before data is our slave and not our master, we may be entering the world of small not big data; small, carefully harvested and curated, and definitely not overwhelming.

Big Data – I reckon – may well be on the way out. Long live small data, with its modest lower case letters.

Watch this space for a forthcoming #SmallDataForum podcast, when Neville Hobson and I discuss these and other issues with @LexisNexisBISUK’s head of global evaluation and insights, @thomasstoeckle1.