Image courtesy of ExasolAG
Customer analytics come from a variety of sources:
- Google Analytics – which becomes more robust every year
- Primary research
- Internal sales databases
- Login databases
- Customer service
- Potentially other sources specific to a particular company
Harvesting this data and successfully utilizing it relies on doing 4 things really well:
- Understanding concepts of consumer (customer) behavior
- Gathering data across functional silos and disparate sources
- Gleaning insights from a blending of data and relevant concepts from consumer behavior
- Implementing change based on your insights
And, based on a survey of marketing managers, alteryx finds more than half feel these challenges keep them from optimizing long-run ROI. So, let’s take a look at each of these challenges and ways to overcome them. See the infographic they produced based on this survey at the bottom of this post.
Customer analytics problem 1: consumer behavior
I’ve been a marketing professor for over 20 years. Most of the schools that I know of require marketing students take a consumer behavior course or, at least, strongly suggests they take one. So, marketing students come out knowing a lot about the decision-making process that ended with consumers either buying or not buying their products. They understand how peers and other influencers, memory and learned behaviors, and cultural beliefs impact this decision-making process. Thus, they know what variables likely impact purchase decisions, so they know which data is important and which has little to no impact on buying decisions.
Even price is a poor predictor of purchase behavior. For instance, Apple sells a ton of PCs, tablets, and other devices despite pricing their products substantially higher than competitors. And, the decisions have little to do with other factors we commonly think of as driving customer purchase, such as quality, availability, etc. And, Apple isn’t the only case where consumers make decisions that don’t fit with our economic notions of what drives behavior.
The problem occurs that these same marketing students who have such a clear grasp of the consumer behavior process as it relates to purchase decisions have poor analytical skills and lack skills in related aspects necessary to derive meaning from data, such as SQL, which we’ll discuss in a few minutes.
The same is true for folks trained in analytics, only in reverse. They’re trained in deriving business intelligence (BI) from data, but, because they have no clue about consumer behavior, they have little clue about what to look for, beyond superficial types of data like demographics, which often explain little of why consumers made specific decisions. Without this information to guide their queries, they’re ill-prepared to develop actionable insights that improve ROI, even in the short-run.
The obvious and most practical solution is to train marketing students more thoroughly in customer analytics. We could think about integrating customer analytics into existing marketing courses but, there are a couple of problems with that solution. First, students self-select marketing, at least in large part, because it’s not reliant on math. A related problem is that most BI courses don’t include enough on customer analytics, instead focusing on finance or operations. A second problem is that there’s already a lot of material in these courses, which we already have problems covering in sufficient depth.
We may have to think about adding an additional required course that teaches marketing students how to derive insights from customer data, including issues of how to manage the data, itself.
Customer analytics problem #2: Gathering data
Here again, we have 2 related problems; 1) generating the right data and 2) gathering data across different functional areas.
Generating the right data: solution
If you don’t have the data in the first place, it’s impossible to generate consumer insights. That means you have to work backward from the insights you hope to generate the data necessary to garner those insights. That might mean using tracking codes across different campaigns and channels so you know which translate most effectively into sales. Tracking codes must record actions within a page, not just page views, which is standard in Google Analytics. Other tools may also be necessary to generate the right data.
Gathering data across functional areas: solution
This is actually a two-fold process. First, analysts must know what data exists across the organization and have access to that data. This is challenging in hierarchical organizations where someone else “owns” the data. For instance, I worked with a glass maker to help improve their ROI. Within a few days, it was obvious that a major problem interfered with optimizing profitability, the marketing people had no idea how much it cost to make any product on their product line. That’s because the acquisition information belonged to the accounting department and the production department controlled information about how much material and labor was used to make a particular piece of glass. Thus, pricing was a guess rather than being based on ABC (activity based costing). The company found it was selling some pieces for very little more than it cost to make them.
The solution is to flatten organizations and transfer data ownership to the organization rather than holding it within a particular functional area.
Customer analytics problem #3: Generating insights
True customer insights come when you blend data and consumer behavior knowledge. You need to know what to look for (not all data is equally valuable). Conversely, you need to understand what the data means by comparing it to what you know about consumer behavior. We already talked about this earlier.
But, often, databases have different formats and a key is required to link databases together. This may mean using SQL to link data from different databases together before running customer analytics to gather insights. Again, most marketing students don’t have this skill set and programs must develop to teach this skill.
Customer analytics problem #4: Implementing change
Implementing change may be the most difficult, time-consuming, and expensive task related to customer analytics. It’s also where improved efficiencies and ROI emerge. Without effective implementation, the data is a meaningless exercise. Again, this is 2 related problems; 1) knowing what the data tells you to do and 2) implementation.
Again, this is 2 related problems; 1) knowing what the data tells you to do and 2) implementation.
Some changes are obvious. If one type of offer works best or one channel seriously outperforms others, the solution to implement change is clear. In most cases, the data don’t tell you exactly what to do. For instance, an increase or an already high level of returns and/ or service calls tells you something is wrong, but a deeper investigation is needed to figure out what change is required. Similarly, finding unmet customer needs suggests introducing new or improved products, but finding these unmet needs is partly data, but more critically, being able to find and interpret statements reflecting unmet needs.
Even once you find opportunities in your metrics, it isn’t always easy to implement change. Sometimes there’s internal resistance to change. For instance, I visited a company who recently implemented a computerized order system (EDI), yet observing operations it was clear that employees didn’t trust the system. Thus, they were still doing manual order entry and processing, using the computerized system to record what they did rather than to optimize order processing as it was intended.
Implementing change may also require some hard choices. Sometimes, it may mean firing long-time employees in favor of new ones with skills that map better to the skill set required after the change.