It’s pretty safe to assume that the Rolling Stones weren’t singing about data when the rock band wrote “You Can’t Always Get What You Want.” But the smash hit’s message rings true for Mick and marketing fans alike. When asking customers for data, you can’t always get what you want. But if you try sometimes, well, you might find you get what you need.
Four data experts break down the do’s and don’ts of how marketers can make the most of the data that’s served to them directly.
Do
Ensure that your data is accurate. When building a comprehensible database, marketers should compile all of their customer data into a single warehouse, says James Kobielus, senior program director of product marketing for Big Data analytics at IBM. He notes that before they consolidate their data, marketers need to match, merge, and cleanse their data, such as through identity resolution. Identity resolution helps marketers determine whether a set of data points pertains to a single person, he explains. Not only do marketers need the right tools to make this happen, but they also need the right help. Kobielus advises businesses to have a data steward who’s responsible for keeping the data clean, concise, and up to date.
Measure revenue and engagement. While profit is the ultimate goal, engagement metrics can be a proxy for long-term success by determining whether consumers like a service and want to continue to use it, says Elena Zheleva, lead data scientist at LivingSocial.
Engage your data scientists. “Instead of giving [data scientists] tasks, involve them in the process,” Zheleva says. “Then work collaboratively on solutions. Marketers and data scientists come from different perspectives—both can bring something to the table.”
Share your failures. No one wants to admit that their data analysis wasn’t as telling as they’d hoped. But Claudia Perlich says sharing failures is crucial to learning. Perlich, chief scientist for marketing technology company Dstillery, also encourages marketers to collaborate with other departments, such as IT, to develop their skill sets and help others understand the problem that the marketing team is trying to solve.
Don’t
Collect data without a purpose. “You shouldn’t be capturing and collecting data without some sense of what you’re going to do with it,” says Jay Henderson, strategy director for IBM Smarter Commerce. Henderson advises marketers to outline their objectives before collecting data and to focus on solving a problem or achieving a business goal. “That often can help make sure that you’re not just reporting facts but that you’re constructing a narrative that’s going to suggest action on the other end,” he says.
Expect the data to interpret itself. “Data itself is inert, [and] it doesn’t say anything,” IBM’s Kobielus says. “You’re trying to tell a fact-based tale. That often starts with conditionals [or] hypotheticals.” In addition to being a great storyteller, two of the most powerful characteristics that analysts can have are good intuition and a strong sense of curiosity, Henderson says.
Improperly set up your A/B tests. When conducting A/B tests, it’s important to compare data samples from the same populations and time periods, says LivingSocial’s Zheleva. For example, when determining a new message’s effectiveness, marketers shouldn’t compare responses from engaged and unengaged consumers. Likewise, when launching a new product, marketers shouldn’t compare yesterday’s metrics to today’s, Zheleva says. Sunday’s metrics may be lower than Monday’s metrics, she notes. Fixate on social sentiment. When analyzed alone, social sentiment fails to tell the entire story of how an audience truly feels, Kobielus says. “[Social sentiment] isn’t really the voice of the people,” he says. “It’s the voice of a very skewed subset of your market.” Many consumers self-censor their posts, he notes, while others only use social media for certain purposes—like work—and don’t reveal their genuine thoughts. So, Kobielus says, marketers should supplement social sentiment with other forms of customer feedback, including surveys and focus groups.
