Although some aspects of marketing will always remain instinctive, each day the field evolves into more an exact discipline—a science if you will. Marketers are continually working with teams of data scientists, constantly tracking actionable metrics, and perpetually creating data-driven campaigns.
Matt Dopkiss, CEO and founder of Dynamit, a data, design, and technology company, is one proponent of scientific marketing. Drawing from his days while pursuing a Ph.D. in physics, Dopkiss says marketers should take a more scientific approach when creating campaigns. He says these methodical rules “provide a wonderful foundation for strategy and produce statistical, measurable results.”
With the constant tracking of metrics, such as KPIs and sales results, Dopkiss says goals can be measured for accuracy in seemingly every marketing channel—social, email, mobile, display ads, and of course, direct mail. Here, the data wrangler provides five scientific methods to design, measure, and test marketing campaigns.
No. 1: Identify a hypothesis
“Every scientific method starts with a well-formed hypothesis,” Dopkiss says. “Before designing, crafting, and testing campaigns, marketers must know what it is that they want to know.” Marketers might want to determine, for example, which offers attract the most customers or identify the best marketing mix for their brand. Dopkiss says once answered, a well-formed question should have great business value, should be specific, measurable, and actionable. “With a good question in mind, we can then create a hypothesis, which we’ll then test through our marketing efforts,” he says.
No. 2: Use a control model
All marketing efforts, according to Dopkiss, should have a control model. He says control models help marketers quantify the results of their campaigns. “The control model actually highlights what would have happened if there wasn’t a campaign at all,” he explains. “In other words, it shows how the market would have responded without the marketing campaigns.” Email marketers, for example, can create a control group by taking a percentage of their subscribers and omitting them from receiving a message or campaign: “Then marketers can just monitor the difference [in their responses].”
No. 3: Use stats to verify a theory
Just like in science, statistics, Dopkiss says, help marketers validate results. “They help quantify error and uncertainty.” But he warns that marketers must be careful to collect enough data so that the stats are meaningful. “If you don’t collect enough data, if your test groups are too small, or if your observations vary widely, you simply won’t be able to make valid conclusions,” Dopkiss says. However, the opposite can also be true; data sets that are too large can wreak havoc on marketers’ efforts to glean actionable metrics or meaningful results: “Bigger data sets bring more noise; look for trends.”
No. 4: Repeat tests
Dopkiss says that for marketers to make assertions about the results of a particular strategy, campaign, or program they have to be able to reproduce their results. “Anyone might be able to get a result one time,” he explains. “You need to be able to reproduce—in other words repeat—the results before we believe them.” He says time of year, weather, or other variables can make a campaign successful, but that same strategy may fall flat in different context.
No. 5: Apply the right technology
Every scientist needs the proper tools, and marketers are no different. Dopkiss says marketers must use the correct technology tools and methods, and have the right mind-set when they extract actionable information from Big Data and evaluate the success of a campaign. “This is where data science comes into play,” Dopkiss says. “We can use these tools to slice and analyze test results to gain insight.”