//instagram.com/p/nLneboIuF4
Thursday evening saw the fifth in our spin off series of Google Firestarters events for the search and performance marketing community, on the theme of data. It’s been eight years or so since Clive Humby (co-founder of DunHumby) first compared data to oil but somehow there remain huge challenges in extracting the full potential that that metaphor promises. So with this event we wanted to explore some of those challenges and also some of the opportunities that we might be missing, particularly in regards to the joining-up of relevant data, real-time insights, use of first-party data and so on. To help us, we had three excellent speakers who gave us three exceptional provocations.
Alistair Dent (Director of Paid Media at Periscopix) titled his talk ‘Big Data Sucks’ (good start), and dived straight into some of the myths surrounding the focus on so-called ‘big data’. Marketers, said Alistair, do not have what might properly be termed as ‘big data’. Most struggle with how to use the (albeit significant and growing) volumes of data they do have access to and so play around with it in order to try and extract value. There was some good parallels with what Alistair was saying with what ‘undercover economist’ Tim Harford had to say recently in the FT (£), namely that with big data we are in danger of making a big mistake. It is, he says, a vague term for a massive phenomenon that has rapidly become an obsession with some, but can be easily over-simplified, misunderstood, or misapplied. The promise of highly accurate outputs (capturing every data point, making old statistical sampling techniques obsolete) can be overrated and create false positives when we are not, after all, collecting every relevant data point or making wrong connections, or putting too much emphasis on correlation and ignoring causation, or when we are using it to make predictions in unstable, changing environments.
Alistair made the point that Search has potentially huge amounts of data associated with it. Ad Words contains 60 columns just for keywords, equivalent to about 1.15 quintillion possible combinations. At 2% error tolerance that’s 23 quadrillion errors (or double the number of ants living on the earth). So we should always start with a hypothesis rather than the data (as one questioner from the audience pointed out, the hypothesis may well be informed by the data, but the point is still a good one). But with search, feedback is close to realtime, we have very few independent variables, and we almost always have a hypothesis. And we can make regular changes easily, and analyse the reactions to those changes. So we can develop real learnings when we focus on the small things. In Alistair’s words, big data sucks, but little data rocks.
//instagram.com/p/nLmE9aouDd
The title of the talk from our next speaker, Ben Chesser the CEO of Footfall 123, was no less provocative: ‘Stop Ignoring the Elephant’. The Elephant in question is the joining up of offline and online. According to ONS, 88% of sales in the UK last Christmas were offline. Yet despite some nice rhetoric around online being a showroom to drive offline sales and stores as a showroom to drive sales online, there has been very little success in understanding the true relationship between the two. Yet the best retailers know that mutichannel customers typically spend multiples (upto 3-4 times as much according to Nordstrom) of a single channel customer’s spend. In attempting to join up customer journeys between virtual and real world too many customers fall off the the grid. Footfall 123 is interesting because it joins up search and other data with in-store transaction data (via customer profiles, loyalty card data, and technology at the tills). This enables a level of granularity in attributing value to specific online channels and activity in driving offline sales, which means better planning, better spending, better results. But in the words of Daniel Lopez (VP at Mango), ‘clicks don’t buy things, customers do’. So the potential is to use such data to understand the pattern of customer behaviour across multiple outlets and online activity. Crikey.
Our final speaker, Rob Jackson (Co-founder of DBi which is now part of the Havas Media Group) titled his talk ‘The Digital Marketing Data Crisis’ and talked about the current inadequacies in both attributing value to specific activity, and also in joining up disparate data points, even those beyond that which typically sits within marketing (like customer service). Separate systems collecting customer data that might be utilised to create better experiences but the difficulty in joining this data up is a major blocker to joining up customer experiences, and very few companies have a true ‘single customer view’. Rob’s team might not have the complete answer to this, but have been working with tag management software, creating a data layer to stitch disparate datapoints together. The potential of consolidating customer profile, behavioural and multichannel data to understand triggers, behaviours and value is huge.
You can see a Storify of the event here, and as always, Scriberia did a great job of visualising the talks – you can see the full visualisation in all it’s glory here. Thanks to everyone that came, to the speakers, and of-course to Google for hosting. Watch this space for news of the next Firestarters event.
Leave a Reply