In the good old days of, say, 20 years ago, the lifetime value of a typical catalog customer was three times the value of a prospect. That meant you could afford to “invest” in first-time buyers-that is, lose money on prospecting-in order to ultimately obtain a loyal and highly profitable buyer file.
Times have changed. Particularly in gift and apparel categories, the lifetime value of a typical customer is only 1.7 times the value of a first-time buyer. That means “investing” in prospecting doesn’t make as much sense. At our company, Good Catalog Co., it’s standard policy to make money on every mailing, whether it’s a prospect mailing or not.
Frankly, that wouldn’t have been possible 20 years ago, or even 10. When it comes to prospecting, the good old days weren’t all that great. We had very few tools that would tell us what or whether a prospect would buy.
This list of 10 best prospecting concepts shows how far we’ve come. The first nine are, in chronological order, what I consider the best ideas of all time (meaning, of course, the last 30 years). The last idea extends into what I think will be the future. When it comes to prospecting, it could be that the good old days are still ahead.
THE TOP 10 1. List rentals. Back in the 1960s, about the only lists available for rental were compiled lists. Prospecting with compiled lists then was like fishing without bait, since they lacked a crucial predictive element: buying behavior.
Someone somewhere came up with the great idea of trading buyer lists. By renting one another’s lists, catalogers, whether they knew it or not, weren’t buying names; they were buying behaviors. That small gesture founded the modern list industry. Once a few dozen catalogs began renting lists, brokers and managers came into the picture to organize and track the orders. Commissions on the names brought revenue to owners and managers, greasing the wheels of list commerce. Third-party list brokers started negotiating fees and rights between list owners, speeding up the process of getting the right list to the right cataloger.
2. Merge/purge technology. Merge/purge didn’t start off as a prospecting tool. Invented in the 1960s, when “prospecting” meant nothing more than mailing a name and address, merge/purge technology enabled catalogers to de-dupe mail files so that they wouldn’t wind up mailing two catalogs to one customer.
At some point, though, someone noticed that those duplicate names-the multis-responded better than the unique names. That was a big aha! It showed that aggregating data about customers-in this case, whether they’d bought by mail in the past, and if so, from whom-could help predict their likelihood to respond. The power of data aggregation remains at the heart of the most advanced prospecting techniques.
3. RFM. This is a huge concept. Essentially, RFM-recency, frequency, and monetary value-is the first and still the best selection technique any cataloger can use in finding customers.
The late catalog fulfillment guru Stan Fenvessy once told me who’d invented RFM, but unfortunately I’ve since forgotten. What I do know is that during the late 1960s or early 1970s, he or she discovered that the three most important variables in predicting response were how recently a person had ordered, how many purchases she had made, and how much money she had spent. When you apply those variables to any mail order list, you’ve made your prospecting chances about as good as they can get.
Since RFM, we’ve had multiregression modeling, CHAID, neural networks, and all kinds of other useful stuff for segmenting a prospect file. RFM, though, is the granddaddy of predictive modeling. It’s as simple as it is powerful. Its inventor should be publicly credited for creating an extraordinary concept. If anyone’s got the name, send it here.
4. The DMA Mail Preference Service. Granted, a “do not mail” file, which is what the Mail Preference Service is, may not strictly fit the definition of a prospecting concept, let alone a great one. But I would argue that it does, and is. It’s an escape valve for truly unhappy customers invaded by unacceptable amounts of mail. It acknowledges a legitimate customer concern, and it siphons off complaints that otherwise could lead to restrictive federal regulations, like those in Europe.
In fact, my only argument is that the Direct Marketing Association makes it too difficult for customers to get on the file. By forcing customers to write in, rather than phone in, the DMA essentially limits the file to the egregiously abused. An easier “opt in” would build good will, inspire loyalty, and easily eliminate a number of unresponsive consumers. And that’s good for prospecting.
5. Overlay data. Without overlay data, most compiled lists would probably be about as targetable as telephone directories. In the early 1980s, service bureaus finally improved the prospecting power of all those names and addresses by enhancing them with demographic, geographic, lifestyle, product, and census data. In a sense, these companies invented large-scale target marketing.
Overlay data also have enabled some large catalogers to “prospect” more effectively within their house files. When you’re the size of $1 billion-plus casual apparel catalogs L.L. Bean and Lands’ End, and your file already includes 50% of all people who buy by mail, you’ve got to add data for effective mailings. List enchancements such as income, household size, leisure activities, and age have helped big mailers eliminate mail waste through better knowledge of their customer base.
6. Mega-merge (or universal merge) technology. Somewhere between the merge/purge technology of the ’60s and the cooperative databases of the ’90s lies the “mega-merge” of the mid-1980s. Basically, mega-merging occurs when a multititle catalog company feeds all its various customer files along with rental lists into a single merge/purge. The aggregated data help mailers determine the best names to mail within their house files and rentals.
Once again, this principle works on the idea that the more data, the better. If you’re Williams-Sonoma, for instance, you can mega-merge your five catalog house lists with all your rental lists, and then see how many Hold Everything names also appear on the lists that Pottery Barn is renting. With that insight, you can figure out which or how many catalogs those customers should receive.
With mega-merge, catalogers can see more clearly which prospects “look” like their customers. It’s been the power behind the growth of large multititle firms. It’s also a great suppression technique. If a name appears only once in a mega-merge of 20 lists, it’s generally not worth mailing.
7. Cooperative databases. Although cooperative databases reached critical mass only a few years ago, at this point they hold information on most mail order buyers in the country. With hundreds of mailers sharing data, cooperative databases enable catalogers to pick prospecting names from a huge universe, and at a lower cost per name than regular rental lists.
Cooperative databases also added purchase behavior to the menu of predictive variables. By comparing your own customers’ buying behavior to that of the cooperative database file, you can find names of other shoppers who behave and purchase like yours. Using this so-called synergy modeling, many catalogers have ended up with powerful prospecting lists.
But cooperative databases aren’t perfect. Because member catalogers supply all their current data, they often end up with lots of dupes in the merge/purge process. Still, co-ops can often find a catalog’s best prospects, suppress the worst names, and help mailers decide whether some of their “inactive” names might be worth prospecting again. It’s the mega-merge concept carried to a logical and productive extreme.
8. Balance, or back-end, modeling. This is one of those intriguing prospecting techniques that really works, although nobody can say exactly why. It’s also a little complicated. Essentially, balance modeling occurs as an optional last step in prospecting through a cooperative database.
Let’s say you have a 50,000-name house file, and you want to mail 75,000 prospect names for a total mailing of 125,000. You rent outside lists, run synergy models, and so on, and gain a total of 75,000 names. After running the names through a merge/purge, you net 50,000 outside names, which when added to your house file gives you 100,000 names to mail. You still need 25,000 names to make your circulation plan. That’s where balance modeling comes in.
Balance modeling uses multiregression, CHAID, and other analytic tools to find those 25,000 additional names within the cooperative database that were not initially chosen for the merge/purge. Those names “balance” out your mailing.
The mysterious part is that, by and large, “balance” names tend to respond every bit as well as the front-end prospects. That’s counterintuitive, since balance names turn up after you’ve done your “best” list selections. It may be that balance names are more recent than front-end names (which, while they may match the traits of your best buyers, aren’t necessarily recent buyers), or it may be that they receive less mail than those “best” names. Whatever the reason, balance modeling works, and it’s a big idea in prospecting. Although it requires an extra step in mail preparation, it’s well worth the effort.
9. Out-of-category modeling. This is a fairly recent prospecting concept that adds a twist to using a cooperative database. It’s simple, it works, and like balance modeling, it’s counterintuitive. Say you’re a gifts cataloger and you want 100,000 prospect names. Normally you’d use synergy modeling to pick those names from among the gifts mailers within the cooperative universe. These prospect names would ideally look like your own customers in terms of buying behavior, demographics, and purchase history.
But with out-of-category modeling, you fish for prospects in different parts of the cooperative universe. You might, for instance, try food categories, or apparel. In each case, you’re likely to find prospects that “look” a lot like your buyers but simply haven’t made many gift purchases. They are big buyers of other merchandise categories, and for that reason alone, they might be likely to respond to a mailing-plus they are underpromoted by gifts mailers.
10. Aggregated specific product purchase data. This is the next Big Idea, though it still has to overcome some fairly substantial technical barriers. In essence, it would allow us to prospect for customers based on actual items purchased. Instead of “knowing” that Customer X spent $250-$500 on shoes at Talbots, you would know that Customer X spent $325 on two pairs of size-seven mid-heel pumps, one red, one blue. You would also know that she bought a $75 tartan umbrella from Brooks Brothers and a $395 beige designer suit from Barrie Pace, size 12.
So far, of course, the list industry isn’t set up to track and submit that data. Every company uses different item codes; we don’t all agree that blue mid-heel pumps should be coded MHP227B. Moreover, it’s not the item select itself that makes the difference in prospecting; it’s the aggregation of the data. With it, you can create a pattern for making specific, reliable purchase predictions. If you’ve ever ordered from online bookseller Amazon.com, you know what I mean. Because of its computerized word-recognition tracking system, Amazon.com analyzes your history each time you make a purchase and suggests what else you might want to buy.
This development, which can be done only in a co-op data environment, could revolutionize prospecting. Eventually, it could even revolutionize catalogs. Think of it: Right now, many catalogs are very general, throwing lots of products out there in the hope that something will be a hit. With aggregated specific product purchase data, we could know with reasonable certainty that a customer is interested only in sterling silver picture frames or wicker furniture. In time, catalogs would become more vertical, more niche-oriented. Customer lists would become smaller and more vertical. And we’d all see much higher response rates.
We’re still some years away from this prospecting advance. But I’ll be watching it develop with interest. If we’ve come this far in 30 years, imagine what the next 30 could bring.