Getting Insights out of Imperfect Data

Your customer database may not be in the shape you'd like - in fact, you may be hard-pressed to call it a database! While it may be difficult to envision getting meaningful insights out of your database in its current form, SegmentologyTM is a great way to learn about your customers in order to ensure that your database is built properly to address data quality and provide meaningful insights. What can you learn from an imperfect database? Let's start with some simple insights.

Source of Revenue / Customer Value
Which customers provide the highest disproportion of revenue? Hint: You likely have a group of about 10% of your customers that account for 40% of your revenue. Conversely, approximately 40% of your customers account for about 10% of your revenue. How much of your marketing is spent on these customers? Are you spending the same on the top 10% as the bottom 40%? Should you spend more? Can you reduce the amount of marketing to the bottom 40%? This information allows you to immediately impact marketing allocation, potentially freeing up budget for other urgent priorities (such as database upgrades) or other marketing programs. While it would be best to test what happens when you don't market to the bottom 40% of your customers that account for 10% of revenue, if your data infrastructure is problematic for testing (or you need to reallocate budget immediately), reducing or eliminating spending to this group is a good first step.

Lifestage
Another type of insight is based on demographics. We've identified 27 major lifestages that occur with some prevalence in the US. We know where they live, their disposable income, what they buy, and common lifestyles. Once your customer base has been analyzed with respect to customer value, it is insightful to understand where the value is coming from, answering the question: "What do these customers look like?" Your customers can be compared to the overall US population with respect to the 27 lifestages (and individual demographics). The result is a demographic profile that allows you to see how your products and services meet the needs of specific audiences. Besides customer penetration, it is also important to identify where your revenue is coming from. Customers are not all equal (as we learn from a Customer Value analysis), and there may be certain lifestages that outshine others in terms of concentration of revenue.

What about the data?
The biggest issue with customer databases today pertains to name and address information. With online forms filled in by customers, phone support customer service representatives, and in-store clerks all hand-entering names, addresses, and cities, it is easy to see how different versions of names and addresses can exist, all pointing to the same individual and household.

While your name and address quality may not be entirely accurate, common fuzzy matching algorithms can identify misspellings and common - but different - spellings and group customers together within a household. This information can then be evaluated to assist in your internal data quality / hygiene process. We recommend testing "off the shelf" standardization tools and match key processes to start to understand how bad "bad data" is for you. This information can be used for summarizing the data required for developing the Customer Value and Lifestage analyses.

Knowing who your big spenders are (and who they are not) allows you to immediately consider budget reallocation, potentially eliminating wasted marketing spend to the lowest-spending groups.

These two "dimensions" can be analyzed without perfect data. These insights can assist with marketing strategy (how should we best go after our prime audience), and tactics (which customers should be de-marketed, how we should develop versioned communications).

Other Insights
While there are other insights that can be generated from your customer data, they are more prone to specific data quality issues, and need to be evaluated separately. One such issue concerns product hierarchy data. If you cannot consistently, reliably, and accurately categorize products (SKUs) for analytic purposes ("Men's Running Shoes", "Best-seller Fiction"), the ability to find insights relating to product purchasing patterns is limited. The same is true for other items such as store locations (for distance calculations). Additionally, without a high degree of comfort in name and address logic, insights relating to the customer purchase cycle are difficult to uncover reliably. The same is true if there is a bias in the length of time that transactions are stored in a production environment. While we would like to have a rich history of transactional data for analytic purposes, many marketing databases restrict the availability of data to only recent activity (12 months, 36 months, etc.).

Conclusion
Customer insights can be developed without having a near-perfect database. Two examples of critical insights (Customer Value, Lifestage) can be derived with an "ugly" name/address file and a transaction file with revenue. This is our starting point for SegmentologyTM, even with well-defined high-quality databases, and there's no reason why you can't generate these insights today.

CAC Group has a QuickStart Program for companies looking to generate insights quickly. This includes householding, analysis, data enhancement, customer valuation, and lifestage insight development. Call us today at (847) 805-9802 or email brukstales@cac-group.com for more information on how this can be applied to your customer database!