Services

Predictive Modeling

At the core of our mathematical marketing services is our model which predicts future buying behavior for each customer.

The first thing we do for a client is analyze their transaction database and score each customer on five attributes:

1. Customer Loyalty (Loyalty Score)
2. Propensity to buy in the near term (Likely Buyer Score)
3. Propensity to buy more of a product already purchased (Purchase Probability Score)
4. Propensity to buy a specific product never bought before (New Item Score)
5. Risk of defection (Risk Score)

We’ve developed proprietary mathematical algorithms, refined over time, to make these predictions and assign these scores.  The algorithms include traditional variables such as recency, frequency, and monetary value as well as other more sophisticated factors including purchase patterns, market basket patterns and changes in purchase velocity.

These predictions are based on past customer purchase transactions.  We’ve found that predictions based on purchase transactions are more accurate than those based on demographic, psychographic, or behaviorial click-through data.  The model becomes more accurate with more transaction history and also more accurate through multiple analysis/campaign cycles.

Affinity Analysis

The affinity analysis is used primarily to develop cross sell campaigns.  The purchase patterns of those who have bought specific products are analyzed and others with similar purchase patterns are identified who have not yet bought those products.  “New Item Scores” are developed for each of these prospects and can be used to segment the target audience by propensity to purchase the targeted items.

Market Segment Analysis

A client’s customer list can be segmented in virtually unlimited ways.  It can be segmented by any of the attributes of the Loyalty Builders’ predictive model, by any attributes of traditional techniques such as RFM, or by other attribute in the client’s database including demographics, service transactions, and web analytic data.  We can help you find new ways of segmenting that will result in more efficient direct marketing campaigns.

Lifetime Value

We predict a lifetime value for each customer based on their past buying behaviors, so marketers can know the net present value of future purchases by a customer.  This allows a marketer to differentiate marketing strategies based on future value.

RFM Optimization

We begin with a standard RFM analysis and apply additional metrics like change in buying velocity or product and product category information for a more sophisticated segmentation.  For example, traditional RFM does not differentiate between those customers who bought fewer products vs. many products although the monetary value is the same.  By adding awareness of number of products bought, segmentation and campaigns can become more meaningful.

“At Risk” Modeling

All customers are analyzed for their potential to defect and a probability score (“Risk Score”) is assigned.  The amount of revenue at risk is also computed. These customers can then be segmented for appropriate offers and win-back campaigns.

Factorial and Fractional Factorial Design

We are firm believers of consistent testing in direct marketing. This testing can be even more powerful using Factorial Design And Fractional Factorial Design  (aka “multivariate testing” or “multivariable testing”). 

Factorial design testing vs. OFAT (one factor at a time) testing will allow:

• faster test cycles as multiple factors can be tested each cycle
• ability to test the relationship between factors
• smaller sample sizes
• lower cost of testing

Fractional Factorial Design allows a marketer to test a greater number of variables than full Factorial Design.

Read more about testing in our Testing Primer. 

Customer Profiling

In the segmenting and targeting process, we can build in factors in addition to the transaction data analysis including demographic, firmographic, psychographic, and behavioral data.  This will allow segmentation and targeting by these other relevant factors. (e.g. male vs. female, geography, size of office).

Target List Optimization

We offer analysis and consulting on finding non-intuitive targets to increase revenue.  For example, many marketers over-market to their “best customer” segments.  We can help marketers find customer segments with greater growth potential than typical “best customer” list segmentation techniques, e.g. mid-tier customers or 1-2 time buyers.

Variable Content Email & Print Optimization

Once you have a targeted list matching the right customers with the right products, or once you have designed your multivariate test, you need to deliver it dynamically by email or print.  We can help make sure the campaign content meshes with the dynamic content capabilities of your email or print provider.

Campaign Results Measurement & Analysis

After any campaign, the data should be refreshed for an in depth look at campaign performance.  This can include offer redemption rates, products purchased, purchases by targeted and non-targeted customers, performance vs. control groups, breakdown of firmographics and demographic attributes, and recommendations on how to improve the next campaign.