Stay marketing-savvy and tech-savvy. Get the latest in martech by subscribing to MarTech Today.
Five indicators of whether you’re doing everything you should with your business’s data
You think you're doing pretty well until a colleague comes back from a conference spouting buzzwords and asking questions. Columnist Jeff Allen gets past the hype and reveals the key questions you should be asking about your company's data practices.
Your boss or colleague has gone on another one of those industry conference trips. You know the scenario: she heads off on a plane and comes back throwing around buzzwords she read in industry literature or heard in cutting-edge sessions.
Now she wants to know the company’s strategy on content optimization or Big Data. Sometimes, it feels as if the boss brought a monkey back with her, and the little rascal just won’t get off your back. As the new pet project, he consumes your resources and chews up the limited time you have to perform your job well.
Senior managers are great at facing these monkeys. When subordinates bring in monkeys they want to stick to their boss’ backs, managers know to direct the conversation so that, ultimately, subordinates take their monkeys back out with them. And, you can, too.
At some point, these conversations are going to happen in your company, if they haven’t already. So, it’s time to prepare — prepare to unbury yourself from a pile of buzzwords by turning them into intelligent and worthwhile questions about digital marketing maturity — questions you can stick to inquirers’ backs as their new pet projects.
How do you assess your organization’s data maturity, and how can you move it forward?
Truth be told, the entire organization should be on board with thinking about and solving pertinent digital maturity issues. Here are five questions you should be considering to help your company take solid steps toward digital maturity.
1. Where are we using learning algorithms?
When programmable thermostats came out, many people were excited to set them at their desired temperature and then forget them. But they soon became frustrated with having to configure new rules every time the weather changed.
Then Nest introduced its learning thermostat, which uses machine learning, a form of artificial intelligence. It learned when changes were needed based on past activities and adjusted temperatures to meet predicted needs. And, though I have to retrain my Nest periodically when it learns bad habits, these learning algorithms save users time, money and frustration, allowing them to focus on more important things.
Likewise, it’s worthwhile to take the first step toward data maturity by assessing where your organization uses learning algorithms. This is because — like programmable thermostats — many organizations still rely on static rules that assume that when X happens, Y should always follow.
In the same way that the Nest removed the need for regularly resetting static rules, learning algorithms bring intuitive benefits to your organization. A learning algorithm can understand, based on trends and new inputs, what changes should be made over time so you don’t have to perform them manually. In this way, you can free your team to focus on more pressing things, like uncovering the intent behind new behaviors to better meet customers’ needs.
2. Where are we using machine learning?
Right now, Carson Wentz is the golden goose of football. After he was drafted by the Philadelphia Eagles, he obtained a four-year deal with a guaranteed $26.67 million and a $17.6 million signing bonus.
In an industry with few guarantees and pay-per-game deals, this agreement showed that the Eagles had a lot of confidence in Carson — and he lived up to the hype. But he wasn’t easy to find. Because he played for a Division II team — North Dakota State University — he wasn’t on scouts’ regular rounds. To uncover this potential, the Eagles first combed through 135 Division I teams, each with hundreds of players available to be drafted. Then, they assessed the Division II teams. And, there, after a tedious search, they discovered Carson Wentz in the long-tail of their data.
Likewise, many organizations have valuable insights hidden in the long-tail of their data. But there is often such a large volume of data that even when staff members know what to look for, it takes enormous effort to dig deep enough to locate valuable insights. So, many organizations only scratch the surface and miss deeply buried gold nuggets.
Machine learning helps to automate this complex undertaking so that once the initial queries are completed, employees can perform the more complex analysis. Some insights can only be revealed by humans, but it’s beneficial to thoroughly prep all the data so they can be more easily surfaced.
3. How are we using data science for segmentation?
Contextual segmentation helps put the right message in front of the right user on the right channel at the right time. Rather than reveal only what you want customers to do, contextual segmentation reveals what they are most likely to do, as well as what their actual needs are at any given point in their journeys.
This outside-in perspective is important in today’s market. Brands must think, “We have this customer in front of us. What can we do for him?” instead of “We have this campaign. We need an audience for it.”
However, scaling this approach for large and complex audiences requires granular segmentation that considers a vast amount of customer data, including past and likely future behaviors from all touch points. Brands must glean spot-on insights from their data about each customer’s real-time wants and needs. Then, they must create and deliver relevant content dynamically to offer superior customer experiences across all devices and channels. And, they must get it right every time, regardless of audience size.
Contextual segmentation — segmenting based on the facts you have and the context surrounding those facts — is a complex science that requires a huge data set, but machine learning can help you execute it.
Then, lookalike modeling allows brands to put gleaned successes to work long-term. Once contextual segmentation and machine learning yield valuable results, brands can locate unreached similar segments that are likely to also convert into high-value customers via the same targeting practices. Using the same targeting methods to acquire and retain potential customers, they create a perpetual conversion loop.
4. How are we using data science to determine Key Performance Indicators (KPIs)?
The path to converting and retaining today’s customers changes and evolves. To stay ahead of the curve, organizational strategy must stay ahead of these changes.
My colleague Bill Ingram, vice president of analytics and social at Adobe, said it best when he told a group of staffers at an internal meeting, “Think about your own score card. Are you sure it’s filled with the metrics that move the needle for your business, or are they just things you heard at a conference, or inherited from your predecessor? … Orient your investment around the things that move the needle in the business — not just on your dashboard.”
Because shifts happen, brands should initiate a never-ending investigative process to uncover timely and relevant KPIs.
The story told in the book, “Moneyball,” and in the film that followed, is an example of why we should be more skeptical of traditional success indicators. The Oakland A’s had a smaller budget than any team in baseball. Still, they wanted to compete and win against the Boston Red Sox and their robust budget.
The A’s couldn’t pay for players with widely accepted indicators of success — high batting averages and stolen bases, for example. So they put data science to work and discovered success indicators could also be found in offensive success metrics — on-base percentage metrics, for instance. They could afford players with high percentages in these areas, and with them, they won.
This exemplifies something that every organization must learn: We should continually question traditional success indicators, our KPIs, and where we can find relevant ones to meet current business goals. Smart organizations know that KPIs are not won in marketing meetings but are surfaced in data analysis using statistics, statistical methods and mathematics that reveal real-time trends and shifts.
To understand your true KPIs, analyze the varied paths customers travel and understand the numerous milestones they encounter in their journeys from brand awareness to purchase, and ultimately, lifetime customer value. Continuous and real-time analysis helps you detect shifts in KPIs so you can adjust your strategy accordingly.
5. Do we let data run free, or is it hyper-governed?
Many analysts take pride in their work and hold their data close as a result, which results in data hoarding. When data hoarders do give others access, it is with stipulations: “I will present only the data I want to, when I want to, and how I want to.”
This behavior reminds me of when my daughter carefully arranged a tray of frosted animal cookies for her cousins to enjoy. She clearly felt immense pride in her beloved tray of cookies. And she had to warm to the idea of handing it over for others’ consumption.
Likewise, many data hoarders innocently deny others access to their beloved data. Too often, they do so because they don’t have time to prep the data or they fear others won’t know how to extract value from it. So, decisions are made throughout the organization without the contextual information necessary to make good ones.
The question to ask in this case is, “How many employees who could make better decisions with data don’t have access to it?” If there are any employees who could be doing their jobs better if they were provided access to available information, it’s time to democratize data access. Today’s self-service and even custom-view data analysis tools allow employees to extract value from data even without extensive training or data-analysis knowledge.
Put the right tools in place to solve the complexity and access issues that impede meeting customers’ needs. Once you have, you’re only left with the cultural barriers — confirmation bias or nefarious data hoarding, for instance. Still, these tools make moot many excuses that justify data hoarding — and so ease the path to the cultural change necessary for data democratization.
In his book, “How Will you Measure your Life?,” Clayton Christensen, who also authored “The Innovator’s Dilemma,” says that he doesn’t want to tell readers what to think but, instead, how to think — giving them the tools to forge their own paths to success.
In the same way, the wonderful thing about data science is that it doesn’t dictate what to think — it offers a model of how to think. Then, you can use it to solve your unique challenges. When team members inquire about your brand’s data science strategies, help them think through these questions to reveal the areas in which your organization can learn and grow.
Some opinions expressed in this article may be those of a guest author and not necessarily MarTech Today. Staff authors are listed here.