Big Data. Big Promise. Big Caution.

Big Data imageBig data claims to be the new salvation for all businesses. Because, we’re told, big data will discover amazing new truths. Time will tell.

But in the meantime, most big promises should also be accompanied by big cautions. Which one’s are most important as we approach big data? Recently, on the Financial Times website, Tim Harford wrote a blog post on the topic: Big Data: are we making a big mistake. It is one of the few really thoughtful big data discussions we’ve come across in a while.

Critically, he notes that a great deal of “big data” is actually “found data”. With found data we don’t know what’s missing, so it can’t deliver conclusive results. If we want conclusive answers, we need to look beyond found data.

On the broader topic, Harford suggests four essential things to remember when analyzing big data:

1. It’s easy to exaggerate found data effectiveness if we talk about the successes but ignore the risk of false positives. He cites Target’s detection of pregnant women here. While we’ve heard plenty about the one woman they successfully identified, we’ve never heard how many women Target mistakenly thought were pregnant. A big oops.
2. Figuring out “correlations” from big data is cheaper than finding causation. But correlation without causation is generally meaningless – and often leads to destructive choices.
3. The reality of sampling bias still matters as much with big data as it ever has.
4. Those who believe numbers can stand-alone ignore the reality that random, unexplained patterns usually outnumber true findings.

We highly recommend this blog post. And will add our two observations:

1. Service providers are reaping huge profit by getting companies to jump into the “big data industry” – including data suppliers, consultancies, analysts, and ad agencies. If you end up amongst the vast sea of big data evangelists, remember that most of them have embraced big data to make money. The valuable few are the ones who’ve found ways to tease unusually valuable insight from the data, not the ones who’ve drank the big data Kool-Aid.

How big are the profits? Recently printed in Fast Company, “the sales of big-data-related products and services grew to more than $18 billion in 2013.”

2. So far, the “learnings” we’ve seen from big data have tended to be tiny factoids that are interesting, but offer little marketing power. Like any data, the only answers you need are the “actionable” ones – answers you can rely on to create profit.

Copyright 2014 – Atomic Direct – All Rights Reserved

About Doug Garnett
Doug Garnett is an expert introducing innovative consumer products and services to market while driving higher return on innovation investment. His career has been spent in innovation and he is the president of Protonik, LLC - an innovation consultancy focused on marketing and innovation. Prior to founding Protonik, he was founder and CEO of ad agency Atomic Direct.

2 Responses to Big Data. Big Promise. Big Caution.

  1. Steve Mintz says:

    Not sure I completely agree with you. Yes, just identifying what appears to be a big data insight is not enough. Neither is partnering with a big data analyst who just gain insights (valid and invalid ones, at that). However, the insights that are found by data scientists and those plying the big data trade need to be matched to a hypothesis that needs to be tested, to separate correlation from causation. Some of the datasets are so disparate that the task of linking to get insight and proper vetting is impossible. Instead the business owner needs to test the validity of the insight to establish correlation. That’s where the value in big data lies. So, I agree to a point that its not just about identifying insights, those who truly understand big data must go through the rigor of turning to big testing to validate that they have stumbled on a big data insight that is indicative of causation and leads to big revenue.

    • Doug Garnett says:

      Thoughtful comment, Steve. Thanks.

      I don’t disagree with your comments – I guess that’s the point about caution. We need to know and recognize the limitations (cautions) of the data and the findings. Then act upon them with that knowledge – generally testing the theories against “sanity” or in the market (assuming it passes the sanity test).

      What I find, though, is that big data is being taken up quite quickly by people who lack the experience, skills, or mindset to perceive what you perceive. Hence, big companies infer far beyond what’s reasonable and decide it’s conclusive and bet the farm on the answer. That farm, of course, quickly goes bottom up.

      And I suppose that’s my point: Let’s take advantage of its strengths and apply appropriate cautions. What’s been interesting to me is my direct marketing colleagues (who have all worked with “big data” for decades) have been quite positive about the post. Because we’ve all learned about the difference between causation and correlation the hard way…in the real world.

      Thanks for the thoughts.



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: