Authored by Mike Cassidy, BloomReach
You could forgive those attending ClickZ Live 2014 in San Francisco if they found themselves feeling a tad overwhelmed as they dashed from presentation to presentation to hear about programmatic marketing, attribution modeling, social metrics, linguistic analysis, and the like.
Digital marketing, which wasn’t even a thing two decades ago, is evolving at warp speed and the tools and techniques that make it possible are coming at marketing professionals at an ever-accelerating pace.
“Whatever practices you’re currently using will not matter in a year or two,” John Battelle, who helps enterprises get their content discovered, told ClickZ Live attendees. He was talking about apps, but he might as well have been talking about digital marketing in general.
Still, leaving the conference with the impression that digital marketers and e-commerce practitioners are destined to drown in a tsunami of technical tools that no one person or team can tame would be to miss one of the strongest themes of the three-day conference.
It turns out that the secret sauce in email campaigns, ad buys, e-commerce efforts, lead generation, and boosting the bottom line is the proper marriage of humans and the machines that can make them better at what they do.
The theme came up again and again in sessions – either indirectly or overtly. Ben Maitland, executive vice president for sales and marketing at B2B marketing firm Multiview attacked the issue head-on in a presentation on the importance of programmatic ad buying.
Yes, programmatic ad buying, which is all about the machine, right? The algorithm knows how to find your customers when they need to be found and how to deliver them just the right message. Maybe, but how does the algorithm know?
“Programmatic should make you better,” he said, during his talk “RTB & Programmatic Buying: How Leading Brands Are Achieving Key Results.” “It shouldn’t replace you.”
Data alone doesn’t do it, Maitland said, turning to a favorite example of author and Harvard Business School professor Clay Christensen. Christensen likes to cite a fast-food chain that wanted to boost its milkshake sales. Maitland said the chain had segmented its customers and was confident that they needed to focus on mothers with children because they fit the profile of milkshake buyers.
Instead of relying solely on data, though, Christensen’s colleagues stood in one of the restaurants, starting when it opened, and observed the behavior of the humans who patronized the place.
“Within five minutes, they saw milkshakes going out of the building, at 6 a.m.,” Maitland said.
Lots of them. And who was buying them? Commuters, commuters who needed a treat to ease the pain and keep them occupied as they fought their way through traffic.
Data is only part of the equation. To build a successful data-aided business, Maitland said, enterprises have to have:
- Good data
- Marketer-friendly technology
- Domain and marketing expertise – you have to know your business, your industry, and your customers
“It’s sad to see people over-trust the technology,” Maitland told me after his talk. “The idea is to slow down and just use it as a tool.'”
Humans will never match algorithms and computers in raw data processing power, but the data sets they produce will always be secondary to human insight, he said.
“It’s going to be the people, the data, all of the above,” Maitland added. “This notion of man and machine just has to be the answer.”
Katie Seegers, a manager with Amazon Local, made a similar point regarding e-mail campaigns. Yes, you can – and should – test everything, she said. Test-generated data is extremely valuable in determining what grabs consumers’ attention, when consumers are most prone to having their attention grabbed, and the like.
“Nothing is too small to test,” Seegers said during her presentation, “Testing & Targeting: A Guide to Building Emails People Want.” “Small things do make a big difference. It doesn’t matter what you’re looking to test – go for it.”
But you can’t leave it at that. You have to think carefully about what you want to know and how that’s best measured, she said. Before you test, you should spell out the metrics you’ll use to decide whether a strategy is working. Seegers, Amazon Local’s email manager, spells out these:
- Develop a hypothesis.
- Determine success metrics (95 percent confidence level that the tested change will increase open rate without hurting revenue).
- Agree on rollout criteria (95 percent confidence level across a given number of geographic regions or test populations).
And you need to weave human insights into the process. Seegers talked about a successful initiative in which Amazon local sent travel offers to customers who had purchased travel books. A logical hunch, but a hunch – one that leveraged purchasing data.
“You can’t lose that human element,” Seegers told me after her talk. “We’ll let the data tell us, but I wouldn’t have figured that out by just letting the data run. We have to find the nuggets. It’s still kind of letting the humans craft it.”
Maybe that seems entirely intuitive: A machine is only as good as the people who built and programmed it. Whether it’s big or small data, the only good data is actionable data. But there are pressures pushing marketers to give in to the machine, to turn everything over to automation
“I think that the seduction,” Maitland told me after his presentation, “is saving money. And scale.”
Humans add insight, but they also add expense and in general investors are interested in holding down costs and increasing revenue wherever possible.
But Maitland himself and many others at ClickZ Live provided a compelling argument that the best way to build a successful enterprise and the best way to increase profits is to marry the strengths of humans and machines.
All of which turns the accelerating explosion of digital marketing tools from an overwhelming array of solutions to an area of nearly unbounded opportunity.
Mike Cassidy is the storyteller for BloomReach, a marketing application company and developer of the personal discovery platform. Reach him at firstname.lastname@example.org; follow him on Twitter at @mikecassidy.