Loyalty Builders turns customer transaction data into marketing information. We analyze transaction data so our customers will have a better understanding of their customers. We help them discover which customers are likely to buy in the near future and what specific products they are likely to buy. We also help them identify potential defectors, profile customer life cycles and forecast revenues. We are a unique new marketing research tool.
Customer satisfaction is a reflection of past experience; it is an attitudinal measure strongly related to a customer's last transaction. Customer loyalty is about future expectations. It is a behavioral construct that measures intent to repurchase.
We do many different kinds standard and custom analyses. Our basic Loyalty Segmentation Analysis scores and segments the customers. A Purchase Probabilities Analysis identifies the four most likely next purchases by a customer and probability for each in the near future. The Subsequent Years Analysis focuses on which products customers buy at various stages in their customer lifetime. An At-risk Analysis identifies those customers with higher probabilities of defecting.
Loyalty Builders offers its product through a software as a service (SaaS) model. Subscription pricing offers convenient monthly payments and ensures customers receive fast access to our solution and computing resources without having to make large upfront costs or major investments in infrastructure. Specific pricing is dependent on the size of the data, number of transactions and customers, and scope of the work. After just a few marketing campaigns, our customers typically see increases in revenue and gross margin that more than offsets the cost of our solution.
There are two required files plus three optional files, described below.
Transaction file (required): This is a record of who bought what, and when. It must include a transaction date, a transaction amount (which may be $0), one or more product identifiers (such as a SKU or product category) and a customer identifier. There should be one record for each transaction.
Products file (required): This file associates a product name with a SKU or product category. Including this file facilitates using the targeting tool and makes the subsequent downloads more understandable. There should be as many records as there are current and discontinued products in the transaction data.
Customer file (optional): For some services (for example email or direct mail) it is necessary to upload a customer file with one record per customer. This file should include email or postal address as appropriate.
Promotions Transactions file (optional): Send one record for each promotion offered to each customer. If a customer is offered ten promotions, then there should be ten records for that customer. Each record should include a description of the promotion matching a record in the prom_desc file.
Promotions Descriptions file (optional): Use this file to describe promotions and coupons which may appear in the transaction record, with one record per promotion.
We recommend tab delimited text files. There should be one record per line. The first record must contain the field names for each column.
Definitely yes. We can also include promotional data and other behavioral data to make targeting as relevant as possible for your business.
Your data is stored, encrpted, in our servers. The servers are in locked and monitored locations. You also have the choice of supplying only customer ID, omitting name, address and other identifiers so viewers have no real access to customer attributes. We are committed to preserving the privacy of your data and have been doing so for our clients since 1999.
With the knowledge you get from our predictive analytics, you can target customers for specific products and promotions, putting an end to "spray and pray" marketing. You will be able to spot potential defectors before they are lost and thus preserve the revenue stream they contribute. You will be able to identify and support promising new sources of revenue among your clients. Your margins will grow. You will become a customer genius because you will know more about your customers than they know themselves.
Anyone who wants to better understand your business and your customers will be able to apply Loyalty Builders information to their job. Typical users include sales people looking for customers who are overdue for a purchase, customer service people checking on the value of a customer who is on the line with a question, marketing people preparing direct mail campaigns, sales managers who oversee a group of resellers, senior management trying to forecast revenues, operations people looking at product trends.
You won’t need to add staff. In fact we believe our analysis will be most helpful if we’re working the experienced people who really understand your business. If you have a market research group, we’ll work with them. If you don’t have one, you can think of Loyalty Builders as an outsourced marketing research group. When our analysis is finished, anyone in your organization who’s been given a password can use your private Loyalty Builders web portal to work with the analysis.
Yes. A customer is a customer, whether he or she is a consumer, a middleman, a manufacturer or a retailer. Customers buy things and in doing so generate accounting records. We work with both B-to-B and B-to-C clients. Companies selling to consumers are often sophisticated marketers who understand the benefits of our analyses intuitively. B-to-B companies might have a steeper learning curve but may benefit from our work even more than a B-to-C company.
Our basic Loyalty Segmentation Analysis is usually done quarterly. Companies in fast-moving industries may need monthly or even weekly analyses. A Purchase Probabilities Analysis is typically done before every large direct marketing campaign to optimize the effort. Subsequent Years Analysis is a yearly project.
We take the transaction data you send us, run our analytics engine and build a marketing database holding all the transactions and loyalty scoring results. We post that database to your private, secure, customized portal on our website. You access this information over the Internet. With this system you get unparalleled customer knowledge and the tools to grow revenue and margins.
Today, Loyalty Segmentation results can be ready for you in five working days. In the future we expect analysis time to drop to a few hours, with results available in a few days. The initial analysis may take a little longer as we work with our clients to "tune" the analysis to their characteristics and needs. A Purchase Probabilities Analysis or Subsequent Years Analysis can be on your web portal in about a week.
15. I think I know who my best customers are. Why do I need a loyalty analysis?
You probably know who your very best and very worst customers are already. The top 5% and bottom 5% are easy to spot. However, it takes a sophisticated analysis to identify and rank the promising customers among the remaining 90%, and it is from that group that the real growth in your company can come. It is also very difficult to predict purchase probabilities, calculate purchase deficits or analyze customer life cycles. A loyalty analysis is much more than just spotting your best customers.
No. Our experience with this analytic approach, plus our broad business, research and marketing knowledge, give us the ability to interpret the analysis most effectively and give you insights you might otherwise miss. By running the analysis on our servers, we are able to tune the programs to your company's particular characteristics and make the most useful interpretation of the results.
Absolutely. If you knew when all of your customers were going to buy again, you would have little need for our services. You will be surprised, however, by the amount of information in the seemingly erratic behavior of your customers.
When we do a Purchase Probability Analysis, we also do a validity test. This test compares our predictions for a defined period in the recent past with actual results for that period. These validity tests consistently prove the accuracy of Loyalty Builders analytic model.
Our Purchase Probabilities Analysis will identify customers with the desired product among their more likely next purchases. Pick customers where the product is their first or second most likely, and where the probability of those purchases is high.
Based on past behavior, each customer has an expected interval between purchases. A customer's purchase delay is the number of purchases that are expected between their most recent purchase and the date of the analysis. For example, customer A ordinarily makes purchases every 2 months. If the analysis is conducted 3 months after their most recent purchase, the purchase delay will be 1.5 (time since last purchase  divided by expected purchase interval ). Purchase delays less than 1.0 are great; less than 2.0 are very good.
There are two kinds of purchase delay. The first is based on a customer's own purchase pattern (PDefC). The second is based on the median time for next purchase for the customer population as a whole (PDefP). PDefP is used strategically to gauge the overall health of a customer population. PDefC is used tactically to spot customers who are deviating from their typical purchase pattern.
Use the Purchase Deficit-Customer index (PDefC) and associated scatter plot, supplied as standard with a Loyalty Segmentation Analysis. See where the top-ranked loyalty group tapers off, typically between 3 and 5. Customers with PDefC greater than that number are likely to defect. Customers with PDefC much closer to 10 may have already defected.
Using PDefC and PDefP together is an effective way to spot defectors. A customer with a low PurDefP and a high PurDefC is in a critical state. This values characterize a customer who has been purchasing at a higher than average rate (hence low PurDefP) but who is now long overdue for their next purchase (hence the high PurDefC). Clearly, for such customers something has changed and they should be contacted.
Use the Loyalty Segmentation Analysis and the scatter plots. Pick customers in the middle or upper loyalty groups with low to average purchase deficit. Look for regions on the scatter plots where groups of customers in a lower loyalty group are sitting in regions normally occupied by customers in one higher loyalty group. Customers who fit these criteria are likely to buy more and thus move to a higher loyalty group.
The objectives are the same: knowledge discovery in databases. The approaches, however, are different. Data mining uses statistics to find patterns of behavior. Loyalty Builders models the customer as a Finite State Machine (FSM). While the FSM may, at times, use conventional data mining statistical techniques to determine model parameters, it goes beyond data mining by organizing and summarizing the results in terms of a customer's current state, a set of possible events that might modify the state (the "inputs") and the set of possible outcomes (the "outputs") predicted by the model for each input and state combination. Unlike data mining which just searches for patterns, the FSM incorporates a bias representing Loyalty Builders' understanding of what constitutes customer loyalty. The combination of this insight or point of view with the organization of the FSM approach is what produces such accurate results.
Unlike data mining or RFM (Recency, Frequency, Monetary value), Loyalty Builders' analysis is based on a complex nonlinear mathematical model of a company’s customers. Technically, the Loyalty Builders customer model is a Finite State Machine in which a customer’s current state and predicted future behavior are based on:
As a result of the level and sophistication of the analysis underlying Loyalty Builders customer model, it has proven highly successful in providing useful and understandable customer segmentations and in predicting future customer purchase behavior. Unlike data mining, the components of the Finite State Machines have defined meanings whose interpretations are clear and useful in understanding the characteristics of the total customer base and the loyal customers it contains.
A linear metric based on a numerical variable will increase or decrease in a constant proportion to a change in that variable. This is not a desirable property for a loyalty measure. For example, when comparing two average customers purchasing in the $2,000 to $4,000 range, a change in purchase amount of $500 might be a significant difference that should be noted. The same difference in amounts for top customers purchasing in the $20,000 to $40,000 range is unlikely to be important.
Transaction data often includes a few extremely large values which will tend to dominate any linear measurement. If the linear metric scale is set to highlight the big differences which are significant for the larger values, important differences for the majority of smaller values are ignored. Conversely, if the scale is chosen to highlight the smaller differences for the majority of values then unimportant differences in the larger numbers dominate the analysis. Clearly a "sliding scale," which recognizes the changing importance of a given incremental change, is more desirable. The Loyalty Builders' Analytic Engine produces nonlinear metrics, a more mathematical term for "sliding scale."