We use cutting edge approaches to database marketing and mining. 

Finding, modeling, testing and validating direct markets – whether it’s for traditional snail mail, email, online, DRTV, etc. – generally falls into one or several of the following categories: 1. Geodemographic Coding & Analysis.  2. Database Modeling & Prediction Using External Databases.  3. Database Analysis Using Multivariate Analytical Techniques (e.g., Factor Analysis, Clustering, Multiple Regression). 

Direct marketing is always a test. Data are appended, extended, and tested to establish predictive response models for roll out. Each roll out into validated market data becomes itself a test for the roll out which follows, thereby providing a continuum for better response to our clients over the long haul.

Cloning: We can match your house file against a national database of 30 million names that are appended with demographic and psychographic information. A regression analysis generates predictive weights for the demographic and psychographic variables that are used to select “clones” from the balance of the national database. The report detailing the demographic and psychographic makeup of your customer base can be further used to shape offers, copy, artwork and the overall selection of markets and market strategies. It will further provide a basis for mining the customer bases of other products in your product line.

Clustering: Regression-based “cloning” is not possible on lists generally available outside of our national database of 30+ million prospects. However, geodemographic clustering – a zip + 4-based demographic coding of American households – can provide a selection screen for rented or compiled lists that will increase response percentages and make your marketing dollars more effective. We will model your customer base to determine which clusters are significantly over represented in your customer list in relation to national averages. Significant clusters can be used to select from large national buyer lists that have been cluster coded. Geodemographic clusters can also generate zip-select or suppression files for use in renting non-cluster coded lists and will increase the overall effectiveness of all advertising campaigns.

Leveraging Demographic Models: Buyers of any one product in your line are prime targets for upsell and cross-sell offers. Once the demographic and psychographic makeup of your customer is established, the relevant demographic information can be appended to the customer lists of other products. This provides a qualifying screen for likely upsell candidates. The two-stage process leverages your database modeling to maximize the return on upsell and cross-sell mailings; to avoid “intuitive” and perhaps ineffective screens and to avoid the cost of appending information that is not predictive.

Finding Unmet Needs: Analyzing the purchase patterns of your existing customers is at the heart of effective data mining. We use sophisticated analytical tools rarely used outside of academic behavioral research. For example, Factor Analysis is a correlational analysis that will show you which products have true predictive affinity for each other. That means that you’ll know which of your other products your customers want, but perhaps haven’t seen yet. You’ll also see where you may be missing sales opportunities and how best to fit customers to products.

Predicting Future Purchases: The same technique that Polk uses to pick “clones” from its national database can be used to determine which appended demographics will predict buyers and non-buyers from any other list – including product-by-product predictions from within your own list. Multiple Regression separates the wheat from the chaff unlike any other selection criteria you may have used. Multiple regression predicts, individually and with statistical confidence, who will buy and who will not – instead of guessing who may be a good prospect based upon fuzzy intuitive similarities.

Penetration Analysis: Standard Industrial Classification, business size, and geographic region will pinpoint statistically significant differences between your current customer makeup and the universe of potential customers. 

Consumer Affinity Clubs: Agent-activated, real time clustering can find upsell opportunities with every new purchase. Database driven and individualized responses to frequent purchasers weds the effectiveness of recency-frequency marketing to the savvy of sophisticated purchase-history factor analysis.

Business To Business: For large enterprise solutions with long purchase cycles, data mining identifies prospective organizations with matching needs and the key decision makers within them. This allows you to maintain contact with these prospects throughout the year, and beyond, with individually targeted mail and email initiatives generated by the analysis of prospect response patterns and ongoing data enhancements.

Mapping: We can provide national maps that can be used for targeted media buying, to drive retail channel decisions and to increase distribution efficiency. Mapping delivers a national perspective of prospects and customers that is hard to achieve by examining lists of zip codes and counts. Database modeling provides prospect counts that can be mapped by the Survey of Consumer Finance and viewed along side customer distribution maps to see where untapped reserves of customers have yet to be effectively mined. Geographic distribution is provided for prospects and customers in three successively abstracted levels of significance.

Distribution: Raw customer and prospect counts are mapped by SCF. These maps usually track population distribution with hot spots in the large metropolitan areas. 

Penetration: Customer and prospect counts are mapped as a percentage of the population. These maps provide an excellent view of market penetration.

Significance: Market Penetration is mapped as a standardized Z-score. These maps provide penetration significance by measuring the national distribution of your product’s market penetration and pointing up those areas where penetration is significantly higher or lower than the norm.