Big data: Does size really matter?
Big data is becoming one big complex. It is the business world’s newest iteration of "keeping up with the Joneses." The fervor for the latest and most expensive collection systems, infrastructure, and PhD-wielding data scientists is rising with the signature speed -- and blindness -- of a race where means are quickly parting ways with ends.
Big data bashers tend to leverage the same criticisms. Big data doesn’t matter unless you know how to use it. Big data is a lengthy, risky investment. Big data reduces people to predictable machines. Big data is invasive, and so forth.
The critiques don’t seem to be resonating with anyone. The competitive winds have picked up, and everyone is trying to reach the same shore first.
But does all data have to be "big"? For businesses ranging from new startups and SMBs to mega corporations, is "small" data sometimes more useful?
Certain questions can only be answered by big data. Target uses big data to identify pregnant customers. Gilt Group creates over 3,000 unique emails for different segments of its customer base. Wal-Mart is trying to auto-generate shopping lists for its consumers. Small data can’t do this effectively.
Big data can replace a lot of guesswork about who customers are and what they might want, but big data cannot, however, replace conversations. To follow and intelligently join conversations about your brand -- particularly on social media sites -- you need "small data."
If big data is the numerical story of spending behaviors, and how to take advantage of them, small data is a glimpse at customer’s thoughts and emotions. Used properly, it is instant feedback that marketers use to slalom between obstacles. Small data lets us stand out while others are focused too heavily on big data. And most importantly, small data lets us talk directly with consumers.
When we use big data, we don’t want our customers to know. No Target shopper wants to know that Target is expecting her baby. With small data, we want to be transparent and responsive. Our "story" becomes not merely what we say, but what we do.
For instance, imagine that a woman named Liz visits Joe’s Coffee Shop in Manhattan. She orders a latte at 8 am, and it tastes awful. She gets on Twitter at 8:15 to complain that her latte at Joe’s Coffee Shop tasted like mud. Now, consider two different ways this could play out:
Case A: Joe’s only collects "big data," so the marketing team is not monitoring social media activity -- a great reservoir of small (and big) data. Occasionally, the marketers work on pumping "Likes" and promoting new summer drinks, but they don’t track the social conversations. Instead, they track how often and on what days customers like Liz come in to buy a latte, so they can target her with fast expiration coupons aimed at turning an extra day of the week into a latte day. So, this team’s big data operation automatically sends Liz a 50 percent off coupon that will expire in two days.
In Case A, Joe’s Coffee has no idea that Liz hated her latte. Instead of winning Liz over, a big data-driven coupon just annoys her. She doesn’t want more of a product that has failed to meet her expectations. She’s been a regular at Joe’s, and now she’s searching for other coffee shops.
Now consider Case B, where small social media data comes into play.
Case B: Joe’s Coffee tracks mentions of the coffee shop across Facebook, Twitter, Google+, and LinkedIn. They see real-time data on whether mentions are positive or negative. So when Liz tweets that her latest latte was awful, one of Joe’s marketers catches the negative mention, reads it, and sends Liz a personal message apologizing and promising that he will look into the issue. He offers her a free latte coupon, making Liz feel like a valued customer, not a forgotten, quantified habit machine.
The marketer notices more complaints than unusual. By reviewing their timing and location, he discovers that people are tweeting and posting complaints within the same two block radius in Manhattan, all near the same Joe’s. He reaches out to more unhappy customers. More importantly, he contacts the store manager to identify the problem. The marketer discovers that a new barista is using the wrong roast during the busy 6 am to 10 am shift.
In Case B, small data allows Joe’s Coffee to quickly pick up a pattern, communicate with customers, mend the damage and address the problem.
This example illustrates the key difference between big data and small data. Big data is for a company’s closed door operations. It’s all about picking up huge patterns and insights that can’t be gleaned from everyday conversation and interaction with customers. The end goal is an automated, profit making process.
Small data, particularly in social media, is for picking up on immediate, actionable patterns that don’t require rocket science to spot, but do require good judgment and fast action to correct. Small data is about finding ways to build loyalty, appreciation and respect. Spontaneity is one of its strengths. Small data catches social cues so marketers can respond like humans. Small data lets a marketer personify his or her brand.
The trick to data, then, is not making it purely colossal and automated, but instead figuring out what data you need to make customers happy and boost revenue. Stop worrying about the number of data servers, and start figuring out what combination of large and small data will make your organization profitable and likeable.
[Image Credit: JD Hancock on Flickr]