Joe Dalton, Chief Product Officer at SmartFocus, gives a step by step guide to creating a robust customer data model
All of us are engaging with businesses across an increasing number of touch points – websites, social media, in-store, mobile and tablets, etc. Regardless of how we engage, we also expect a customized, personalized, and consistent experience.
This expectation continues to be a challenge for businesses, which have to manipulate enormous amounts of data to try to understand how to effectively engage each individual.
Here are a few steps to help get your marketing data house in order, which is the prerequisite for engaging with each of your customers successfully.
Step 1: Create a Marketing Data Inventory
First, create an inventory of all of your data sources and data producing applications that contribute to your customers’ identity, behavior, and activity profile. Typically this involves purchase history data, CRM data, marketing activity data, ecommerce data, any purchased or 3rd party data, social data, review data, etc.
Once you have an inventory off all of the data you could pull together into one system, rank each source of data by the effort it would take to access it and either copy or index into your marketing data store (depending on how you choose to access it).
Be sure to include data from systems and projects that are currently planned or in-progress, such as in-store beacon pilots, customer loyalty projects, etc. You’ll want to keep a 6-12 month horizon view of what customer and marketing data is coming online. This should result in a table that looks something like this:
Another way to think of this is to create a customer-centric diagram, outlining all of the possible touch points with which your customers interact, and all of the data your customers create through those interactions.
Step 2: Audience Targeting (Segmentation)
Most companies perform some sort of customer segmentation, usually as part of one or more marketing applications currently in use. Email marketing systems are a good example of this. However these systems often fall short of what the company needs to holistically improve its marketing effectiveness.
How should we be segmenting our customer base? How many segments can we effectively market to? How can we track our marketing effectiveness by segment? How do new customers respond to our marketing efforts compared to our loyal customers? How can we leverage historical behavior/purchase data? These questions are usually raised by marketers and analysts when planning a more robust customer analytics and segmentation investment.
This is usually the best place to make an incremental investment in customer analytics for the simple reason that a good foundation in customer discovery and segmentation capabilities helps improve every follow-on or downstream application investment.
Whether you choose to tackle this step with an off-the-shelf application or an in-house application project, the end result should be a list of customer segment definitions that the whole company can agree on and that you will measure over time. What I’ve seen work very well here is a simple scorecard that measures certain key performance metrics per segment and sets a goal for improving those metrics over a defined timeframe. Here’s an example:
Note that you don’t need to have a very big percentage improvement in any of the metrics to see positive results. Usually a 1-2% improvement in any of the metrics will result in measureable bottom line movement.
Step 3: Marketing Attribution
At this point, having invested in a holistic marketing data environment as well as a robust customer analytics and segmentation application, we are ready to tackle more analytically intense projects and applications. Having an understanding of what marketing channels work better than others for each of your customer marketing segments is a good next step.
The primary objective of a good marketing attribution application is very straightforward: help us spend more on what works and less on what doesn’t. There are many attribution methods in use today, but the concept behind all of them is the same: attribute each transaction to the marketing events that preceded that transaction according to certain rules. The most basic of these is either “last click” or “first click”, which gives credit for the entire transaction accordingly.
Equal Weight and Fractional models provide a better picture of what’s going on and are relatively straightforward to implement. A more flexible attribution model will also allow you to look at attribution by customer segment, so you can determine which marketing channels work well for each segment, and which don’t.
In your own company, you might discover that certain marketing channels, like email, work well for older demographic segments but not so well for younger customers. Marketing attribution can help you dial in the right marketing “mix” for each of your segments.
Step 4: Predictive Marketing
The availability of rich data sources as well as a number of options for off-the-shelf applications has significantly lowered the cost of deploying predictive modeling systems, and thus have raised their ROI considerably.
One of the most effective initial use cases to tackle in this area is “propensity to purchase”. For a multi-department retailer, for example, giving a marketer the power to select customers who are likely to purchase from a certain department within a certain timeframe is very helpful in optimizing marketing spend. Deploying a propensity to purchase application that allows the marketer to select a group of people who have a purchase propensity score of between 50% an 80% results in much more targeted and effective marketing budget allocations.
Note that as your organization completes each “step” above, you move further toward the ultimate goal of providing only relevant and timely content and marketing messages to each of your prospects and customers. Building your marketing strategy on a solid customer data foundation will pay dividends for years to come.