Catalog and Direct Selling: Track Retail Results
In the mid 1990s, most direct marketers captured about 80 percent of their orders with source codes. What a difference a decade makes! Today, if multichannel direct marketers capture 50 percent of their orders with source codes, they are doing well.
Tracking response to a direct response promotion in a retail environment can be particularly challenging. Barcodes certainly help. When you mail a postcard that’s also the response device, e.g., “Bring this postcard to your nearest store to receive your discount,” you can track a portion of your orders with a high level of accuracy.
However, any direct marketer who has tried it knows the process isn’t so easy. Broken point-of-sale (POS) scanners, smudged barcodes, and inadequate training of temporary and seasonal staff conspire to turn your well-thought-out plans into results that are far from the clean and organized keycode reports you envisioned.
Unfortunately some direct marketers have given up on the challenge all together. They either think it’s an impossible task, or they mistakenly assume that because their response is occurring within their retail stores the orders are somehow being driven by the order channel—the store—and not the promotional efforts. When these marketers cut mail quantity, they find themselves scrambling to meet projections.
Fortunately, there is help. It’s called a data matchback. Here are seven commonly asked questions about data matchbacks and how this technique can help you track retail customers.
1. What is a matchback?
A matchback is a technique where you take your untracked orders, regardless of source (retail store, phone, Internet, fax), and match them back to your actual mail file, including rented prospect names. A good matchback uses a matchkey strategy that compares only portions of different fields of a customer’s total address. Since keying errors can occur when customers’ addresses are entered into a POS system, it’s more likely that a portion of each field—usually the first two or three digits—is more accurate than the rest of the field. Add a few of these portions of fields together, and you have a matchkey.
Using a matchkey increases the chances of finding a customer in your mail file, compared to using an exact match of name and address. Once you’ve located the customer in question, you can jot down that customer’s source code from the mail file and go back to your records to see what segment or prospect list correlates to the code.
That’s the basic theory of a matchback. But you automate the process and run thousands of orders in a pass.
2. Is a matchback better than a percentage-based reallocation process?
Yes. Much better. But first, what is a reallocation?
When direct marketers have a large number of orders but can’t run a matchback, a standard practice is to reallocate the unmatched orders by the tracked response. If a list or segment receives 5 percent of the tracked orders, then 5 percent of the untracked orders are assigned to it.
In a retail environment, marketers can isolate orders by ZIP codes within a radius of a store. This can improve the results in the reallocation process. However, customers don’t necessarily take into account the proximity of the nearest store to their home when they shop. They may go to a more distant store because it’s on their way home from work.
At times, a reallocation can be better than nothing. However, there are always two questions. First: Did the orders really come from these lists and the direct mail promotion? Second, even if you can isolate the orders driven by direct mail promotion, does a simple reallocation really hold up on a segment-by-segment basis?
For example, in one recent head-to-head test between a reallocation and a matchback, the two techniques produced results within plus or minus 7 percent of each other on the housefile segments.
The prospect lists were a different story. The matchback increased all of the prospect list responses by an average of more than 20 percent. One list’s response increased 50 percent. For several of the rented files, that was the difference between success and failure.
The lesson? If you have the task of growing your company’s sales, a matchback can be essential to determine which of your prospecting efforts really are working.
3. Are these matchback orders really from the people we mailed?
For smaller mailers, spot-checking the names and addresses of orders that match those of the mail file is simple. You literally can lay the names and addresses side by side and compare them.
For larger mailers, a more effective technique is to run a second match. Take your matched orders and tighten up the match criteria against the matched portion of the mail file. Now visually check any orders that don’t pass the new, more stringent criteria. Were these matches rejected due to simple misspellings or variations such as the difference between the words “street” and “avenue?” Or is it obvious that the match is not correct?
If you see significant problems, you may have to run a multitiered matchback in which you progressively tighten your criteria to eliminate errors.
4. If we re-mail the same lists several times, how do we know which mailing produced the order?
For prospecting files, this usually is a straightforward question. You often rent and mail the same name once during a season. If you match an order to the name, bingo.
Housefiles and rented multibuyers that receive two or more re-mails are a different story. In this case, it helps to plan ahead. For example, if you assign different offers to the various drops, you often can isolate many of your orders by the offer.
However, at this stage of the process, you usually have to make some educated assumptions. If, for example, you’re a holiday mailer and you sent four promotional pieces to your top house segments from September to December, it’s reasonable to use the order date as the assignment criteria. However, your results now are less precise.
Here are two ways to crosscheck your results:
• Use your order curve. Plot your daily order curve and identify your drop dates in the flow. Compare the spikes in orders to the number of orders assigned to the re-mails. Does the correlation of the matchback orders, the drop dates and the order curve make sense?
• Use control groups. If you have large enough segments, maintain control groups that receive different numbers of mailings. You now can track the lift that your re-mails produce in the test segments. Specifically, do the segment penetration rates (total number of orders for unique names) correlate with your matchback results?
If you’re a multichannel marketer with retail locations, control groups should be an essential part of your direct mail planning. For example, one multichannel marketer regularly selects a control group that doesn’t receive an upcoming promotion sent to its housefile. Though it still struggles with tracking response, by holding out a control group it has determined that receiving a catalog mailing produces, on average, a 20 percent increase in its customers’ purchases.
That’s not enough information to determine which customers to mail and how frequently. However, it’s enough data to know that its catalog program and its response tracking systems are worth some investment.
5. Can you run a matchback in-house?
Yes, a person with a working knowledge of Microsoft Access and an understanding of matchkey logic can run a matchback in-house. However, if you’re planning to run a matchback for prospect names, you’ll need to obtain permission up front from all of the various list owners. When you request permission, you have to disclose that you are running the matchback in-house.
If the list owner refuses that request, it usually will allow a recognized third party, such as your service bureau, your agency or a company recommended by one of them, to run the match.
Regardless of who runs the match, you’ll need to arrange for your service bureau or agency to save a copy of the mail file. You may obtain permission from list owners to run the matchback, but without a copy of the mail file, you won’t get far in the process.
6. What match rates can you expect?
The percentage of orders that can be matched varies widely. The lowest match rates usually hover around 35 percent of the unmatched orders. If you have good accuracy at your POS level or call center, your match rate often can top 55 percent. It’s not uncommon to match 70 percent or even 80 percent of the originally unmatched orders.
7. Does a matchback really make a difference?
Yes. Of all of the marketing disciplines, direct marketing has the advantage of being the most predictive. We can track results and use real data to forecast projections. When our colleagues from the mass marketing side of our profession see the level of accuracy of our forecasts and the precision of our tests, they usually are envious of this predictive quality. However, when we don’t track results, we lose our primary advantage.
Though it may be a new technique to you, start testing and taking advantage of matchbacks in your marketing efforts. Improved tracking and reporting leads to every direct marketer’s ultimate goal: improved results.
George Hague is the senior marketing strategist at J. Schmid & Associates, a catalog consulting firm in Mission, Kan. You can reach him at (913) 236-8988 or by e-mail at firstname.lastname@example.org.
A columnist for Retail Online Integration, George founded HAGUEdirect, a marketing agency. Previously he was a member of the Shawnee Mission, Kan.-based consulting and creative agency J. Schmid & Assoc. He has more than 10 years of experience in circulation, advertising, consulting and financial strategy in the catalog/retail industry. George's expertise includes circulation strategy, mailing execution, response analysis and financial planning. Before joining J. Schmid, George worked as catalog marketing director at Dynamic Resource Group, where he was responsible for marketing and merchandising for the Annie's Attic Needlecraft catalog, the Clotilde Sewing Notions catalog, the House of White Birches Quilter's catalog and three book clubs. George also worked on corporate acquisitions.