For marketers who provide seasonal products, driving off-season sales is often challenging. However, by using predictive analytics, Billie Phillips, vice president of marketing at the Salem, Mass.-based candy company Harbor Sweets, was able to identify “at-risk” customers and market to them without risking valuable marketing budget dollars. Phillips used Portsmouth, N.H.-based Loyalty Builders’ Longbow analytics software to analyze customers with an order history that indicated they were buying less frequently and then apply this insight to a targeted mail campaign. With a response rate of nearly 40 percent, Harbor Sweets was able to increase results from a market that, previously, the company was slowly losing touch with.
Target Marketing: How does predictive analytics allow you to find and reach out to at-risk customers?
Billie Phillips: We recently analyzed a number of customers that looked like they were at risk—they had not ordered for a while—but had a good loyalty scoring and had been interested customers when they ordered in the past. We determined how many of those names we wanted to contact, which allowed us to stay within a budget we could control. Then, we mailed a personal letter to them, and shortly after the mailing, we had a 9 percent response rate. The rate then grew, over time, to almost 40 percent.
In terms of the offer, the added benefit was that it made people remember Harbor Sweets. So the residual response grew, and the people who got that letter continued to order through the summer and fall. The interesting thing was [that] the special we offered them was good through Mother’s Day, and while the entire group did not take advantage of the initial offer, there was a response that continued into the summer and the fall.
TM: How did the new approach improve your ability to effectively allocate marketing dollars?
BP: The added element that enabled us to predict how customers are going to respond made a world of difference in our campaign preparation. In addition, before you have committed to a particular promotional campaign, the program allows you to estimate what your response is going to be in a dollar value. This was particularly useful because it helped us to stay within our marketing budget. We can now say, “We are going to spend X amount of dollars to get to this group of people,” then quantify it so you can get a sense of what your ROI will be. For us, while it was not exact, it was very close to what Longbow’s results told me we should anticipate for the effort. The key is that you can be far more efficient with your marketing dollars … especially in nonholiday seasons.
TM: What were the results of your recent suppression test to assess how effective the annual holiday catalog mailings were in driving sales?
BP: The results of that suppression test showed us that we would achieve a boost in our sales if we were to redistribute the number of catalogs we are sending, and if we stay in close communication with those people who are repeat buyers. The question we asked was, “How does our mailing schedule affect those best buyers; what would happen if we didn’t send the number of catalogs we send through the fall; what if we redistributed to another segment of the list?” We did a very small sampling the year before this past holiday which indicated that we may see a positive outcome by mailing less catalogs to that group—it was really a test for us—and during the second test, the results were the same. So, we now know that we could redistribute and see a lift in the sales instead of losing sales. That, combined with the ability to reach at-risk customers, has improved our timing, and when mailing, timing is everything.
- Companies:
- Harbor Sweets
- Longbow
- Target
- Places:
- Portsmouth, N.H.
- Salem, Mass.





