Today’s column is written by Lauren Moores, vice president of analytics at Dstillery.
Big data is one of those wonderful concepts meant to make life easier and business more efficient.
But in practice, before we get to any of its promised benefits, it makes everything more challenging. We have so much data to work with that we don’t know which questions to ask, much less which to answer.
And herein lies the issue. If you cannot discern which data is meaningful and what queries to apply, you can construct all the algorithms in the world to drive programmatic decision-making, but the results will be less than satisfactory. There is no simple solution to this conundrum, but there are ways to steer your data in the right direction. The key lies in data relevance.
I look at big data in the context of my personal experience and how today's data compares to yesterday's. When I was creating economic forecasts for various industry sectors, we used time series data, drawn from many years at monthly intervals. Processed on big Honeywell machines, accessed through TI thermal paper terminals, we used a proprietary language to build metadata for use in analyses.
Fast forward to the early online bubble. We used multiple data sources to achieve volume and scale of data that enabled us to rank and use only the quality monthly data in our company information products. Megabytes became gigabytes, and soon, if you weren't working with terabytes of real-time data for weekly and monthly decision-making, you were missing consumer behavior signals. Today, terabytes and petabytes are the norm but, more importantly, real-time decision-making is also the standard practice.
Data And The America’s Cup
Think about it. “Wanted: Ability to handle large amounts of data streaming in real time; ability to create patterns from thousands upon thousands of parameters; ability to build agile algorithms to handle data programmatically.”
Many of us reading this want ad would think it was just another search for a data scientist at an advertising or media firm. However, having moved from academia to finance to advertising, I have observed the building of programmatic solutions using big data become the “in” strategy for all industries.
The want ad above is a hypothetical description for the America's Cup AC72 analyst role. Sailing aficionados may focus on the AC72 design – lacking a true hull, outfitted with a 130 wing sail, use of carbon fiber and titanium and built to hydrofoil – but data geeks everywhere are talking about the data that went into this victory. With more analysts than sailors, programmatic big data analyses allowed the US team to make minor adjustments and tweaks in real time and gain the critical advantage needed to defend a trophy that New Zealand appeared to have already won.
In all of my experience with big data in a variety of industries, the key is to understand which data is relevant and which is not. Real-time decision-making fails when the data produces noise rather than signal. The old expression, “garbage in, garbage out,” holds here.
In the world of advertising, it’s particularly challenging. We have so much data about prospective customers that it’s difficult to determine which are relevant to our targeting, modeling and CRM efforts. Furthermore, we rightly endeavor to create multichannel and multiscreen advertising, but each campaign and screen requires different targeting criteria, with different data-based success metrics to measure success.
With all of this complexity, how can marketing leaders use data to help make better decisions about their businesses? What’s the best way to organize and align data with business objectives to determine strengths and weaknesses, and to plot strategy?
There’s no easy answer. There is, however, constantly evolving and improving technology to help decode the data and make the process simpler. Using the right technology platforms to help untangle the massive and unruly web of data is the best first step. As with the America’s Cup analysts, it’s all about understanding the goals and mapping the relevant data back to them. If you’re confident in your business objectives and KPIs (winning the race) and conscious of your obstacles (wind, currents, competitors), it’s easier to determine which data is relevant.
“Wanted: Ability to handle large amounts of data streaming in real time; ability to create patterns from thousands upon thousands of parameters; ability to build agile algorithms to handle data programmatically.”
For marketers, the search is over.
Email This Post