When it comes to mailing list services and autoresponders, few services out there are more popular than MailChimp. As of June 2014, MailChimp was sending out over 10 billion emails a month. That’s a lot of communication. And there’s no one better suited to wrangle that communication than MailChimp data scientist and author, John Foreman.
His insights and the data science work he’s done for MailChimp, as well as other companies like Coca-Cola, Royal Caribbean, Dell and more have helped not only grow these respective businesses, but also forged new paths in understanding what it means to create impactful, actionable and meaningful data science discoveries.
So how can you follow his lead and put data science to work for you? Let’s take a closer look:
First, Ask “What’s Useful or Required by the Customer?”
One of the biggest misconceptions about data science is that there are a bunch of number-crunchers holed up in their silo cubicles, staring at mounds and mounds of spreadsheets and screens and having little to no interaction with anyone else. According to Foreman, nothing could be farther from the truth. As the brains behind many of the data-backed initiatives at MailChimp, Foreman and his department serve teams internally as well as MailChimp’s customer base.
One such example of an internally-developed tool that was eventually rolled out to the mail-sending population at large within MailChimp was Omnivore, an artificial intelligence learning tool that scans emails for bad URLs, spammy keywords and other telltale signs of mail abuse. It is, essentially, a self-cleaning tool that continually adapts to not just help ban abusers from compromising the system, but helps keep inboxes as a whole, cleaner and safer.
Tools like this one are not only useful, they’re necessary for a company like MailChimp, whose very business backbone is built on the ability to send clean, reputable messages. One thing Foreman adamantly wants to avoid is the perception that data is glitzy, glamorous or worth showing off just because it’s the trendy thing to do.
According to an interview in GNIP, he stated “[a] data science team is not a research group at a university, nor is it a place just to show off technologies to investors. We’re not here to publish, and we’re not here to build ‘look at our data…oooo’ products for the media.” He further adds, “whenever a data science team is involved in those activities, assume the business doesn’t actually know what to do with the technical resources they’ve hired.”
In another, meatier example, profiled on Mashable, the issue at hand was that ever-present, ever-nagging question:
“When is the Best Time to Send Email?”
Foreman has stated that self-doubt holds back users from achieving the kind of success they want in MailChimp, so the team built a Send-Time Optimization System (STO) to mine the data. Here’s what they got when they segmented recipients into three groups: college-age, forties and over-retirement age.
So there is, as you might expect, no “one size fits all” approach to the absolute “best time to send” email messages. It depends on whom you’re targeting: not just their age group, but also other demographics like their work, where they live, and so on. If you’re wondering why they built an entire system just to prove what most marketers know as common sense, the answer is exactly that.
The data proves common sense, and MailChimp gets a precise, data-backed answer that isn’t just pulled out of a customer support technician’s ear. This, in turn helps customers feel more confident about their segmentations and get better results from their emails, which in turn helps fuel their connection with and loyalty to the MailChimp service.
Circumvent Unnecessary Risk and Complexity
Another surprising bit of information, which may come as a surprise to some marketers who dabble in development, is that sometimes the simplest option is the best.
Foreman was able to demonstrate his expertise to the MailChimp team when he originally applied for the job, not by using flashy platforms like Hadoop or NoSQL, but by using plain old Excel spreadsheets. He follows this same path in his book, as well.
According to an article in VentureBeat, Foreman notes, “a data science team should align itself with the business and serve that business…[t]he purpose of the data science team is to lead from the back, not to make headlines.” And one of the many ways they do that is by avoiding using flashy new technology just because it’s new. You end up avoiding a lot of unnecessary complexity that way.
Data Science is Not Just “Nice to Have” – It’s Necessary.
As more and more companies start to gather meaningful, usable data, the science and the users behind it become even more critical. Having “data backed” hard evidence that you can show customers, team members and investor alike helps breed confidence, and confidence breeds success.
All of this ties in with technology in that we’re getting both a broader picture and a granular view of how our customer base is acting and reacting. We can see how data overlaps and bridges gaps while zooming in on a single point to get the kind of details that would ordinarily be pure guesswork.
Are you collecting the kind of meaningful, measurable data that can make these types of impacts in your own work? While you may not have the flexibility to create entire A.I. systems just to prove or disprove a point, the fact is that having data on your side solidifies your company’s place in its respective industry and in the small but measurable things you do to improve the customer experience.
How are you using the data you collect? Have you been able to forge some new ideas or hypotheses to test? What have the results been? Share your thoughts and perspectives with us in the comments below.
About the Author: Sherice Jacob helps business owners improve website design and increase conversion rates through compelling copywriting, user-friendly design and smart analytics analysis. Learn more at iElectrify.com and download your free web copy tune-up and conversion checklist today!