Special Report Matchbacks
By Terrell Sellix
A matchback is the process of matching order records back to mailing-tape records to determine the actual source of those orders. Matchbacks have been used for years on a limited basis to try to pinpoint the source of unknown orders: typically 5 percent to 20 percent of orders. With the advent of the Web and the increase in multichannel marketing, understanding where your orders and customers are coming from has become harder to learn — and yet more critical to know — than ever.
The shift has brought matchbacks into the limelight of customer order-tracking and results analysis. This Special Report will outline for you:
- how multichannel marketers are using matchback techniques today;
- steps to take to perform expert matchbacks; and
- techniques and tools to get matchbacks done in a way that's best for you.
Eight Lessons You Can Learn From a Matchback
As a direct marketer, you have the advantage of measuring your successes (and unfortunately sometimes your mistakes) in ways that general advertising cannot. You meticulously test, code, track and analyze the results of your prospecting efforts. Such tactics have generated accurate metrics that helped guide you in meaningful directions. At least until recently.
Today, knowing from where your orders and customers hail has become increasingly difficult. It's the rare direct marketer who can survive in a single marketing channel, and most have at least two channels: catalog and Internet. Add retail locations, special mailings and opt-in e-mail campaigns, and the task of tracking sales and allocating marketing dollars has become even more challenging.
That's why the topic of matchbacks has been such a hot one in recent years. Simply stated, matchbacks have brought back your ability to connect marketing dollars with marketing results. Or in the case of larger companies with additional marketing channels (e.g., brick and mortar stores), sophisticated solutions have, for the first time, enabled the connection of non-direct revenue with direct-selling channels. This connection gives you the unprecedented ability to effectively allocate resources across channels.
Most matchback reports aren't any different from normal keycode results reporting. They match orders with no keycode to names from the mail tapes whose keycodes you can identify. Once the name is identified, the appropriate keycode can be assigned to each previously unknown order source, and overall results for that list can be analyzed. See Simplified Matchback Report.
|Pre-matchback Sales||Unknown Sales Matched Back to Code||Total Sales Allocated to Source|
|House list A||$4,000||$1,400||$5,400|
|Prospect list B||$2,271||$455||$2,726|
|Prospect list C||$1,906||$1,100||$3,006|
Prior to the matchback, 31.4 percent of this merchant's sales were from unknown sources. After the matchback, the proportion of unknown sales was reduced to 6.7 percent. This cataloger now can make informed decisions about where to allocate funds by looking at the detailed keycode results available.
Here's what you can learn from looking at your matchback results.
1. Appropriate resource allocation by channel. When the Internet first took off, many companies thought customers eventually would stop ordering from the printed catalog and move exclusively to online ordering. And at first it appeared that the trend just might go that way as Web sales began to comprise 10 percent, then 20 percent, even 40 percent or more of total sales.
However, matchbacks have provided the proof that many catalogers needed to justify resource allocation to their marketing programs, as they show Web sales correspond directly to mailed catalogs. It's accepted now that for companies that mail a printed catalog, typically 80 percent of their Web demand is generated by their catalog mailings. One has only to look at order curves on online channels to see that online order activity peaks around catalog in-home dates.
2. Statistical significance of your results. Did you receive enough responses to statistically prove that a result isn't an anomaly but rather is representative of a pattern? Catalogers usually prefer to see at least 50 responses (orders) for a given group to say those responses are statistically significant. By performing a matchback, you increase the likelihood that groups will reach that minimum 50 responses, thereby enabling you to conclude statistically significant response patterns for each group.
3. List-rental results. Perhaps the most common reaction we get after performing a matchback for our catalog clients is: "Wow, I was going to omit that list from future mailings. It turns out instead that it's one of my top-performing lists." Matchbacks help you zero in on the best, most profitable lists.
4. The best way to optimize housefile mailings. It's common practice to mail housefile names even if they haven't purchased within the last 24 months. However, as those names get older on your file, the frequency with which they should be mailed declines. By bringing real results back to the table, matchbacks help you identify the best mailing frequency. In addition, they help you identify how deeply you can mail into older-name segments and determine which segments may be better off going into the pool for an optimization model with one or more of the co-op databases.
5. The best co-op model selection. The old adage, "Don't put all your eggs in one basket," is an appropriate one for catalogers, particularly in the test stages of any mail campaign. Matchbacks help you to know which baskets to focus on by showing you where your investment delivered the best results. This is particularly important with co-ops where large universes of names are available if results prove promising.
6. How to create offers that pull responses. Matchbacks give you the raw data to accurately assess online order behavior based on target groups. These data can help you create catalog and online offers that target specific groups based on their buying behaviors. You're able to look at key metrics such as: recency, frequency and monetary value per online buying group; how average order sizes differ across channels; and what customers are ordering from multiple channels.
7. Identify core customers. Many studies have shown that consumers who buy from more than one channel are more valuable than those who buy from a single channel. Matchbacks enable you to identify these multichannel customers and therefore target your contact strategy accordingly.
8. Track catalog tails. Sometimes it's useful to match back orders even when a keycode was given. I'm continually amazed to find orders that came in during the month of, say, September with codes that were mailed last November.
Many experts note that a successful matchback rate must bring your percentage of unknown sales to 15 percent or less. But depending on your total sales, a successful matchback rate is anything that's strategically and statistically useful. Typically, we like to see at least 65 percent of unknowns/Internet sales matched back to mailed codes.
From general decisions about what marketing channels to invest in, to detailed decisions about what outside lists to rent for specific mailings, matchbacks give you back the ability to know where to put your money in order to maximize results.
Matchback Do's and Don'ts
The right steps for successful matchbacks.
The right steps for successful matchbacks differ only in their complexity of rules, priorities and match criteria. Not surprisingly, the rules applied get increasingly more complex the more contact pieces you have in the mail at one time. Therefore, well thought-out criteria are critical in maximizing the usefulness of the matchback process.
Do gather relevant data from your matchback vendor. Include in your file appropriate sales data such as name, address and customer number. Also include all order information such as date of purchase; order total with or without shipping (depending on how you typically do your reporting); and product information if you plan to analyze post-matchback product sales.
If you're using standard order export technology (e.g., Dydacomp's Mail Order Manager Export Module), do tell your vendor which fields to include in the matchback and which to ignore. And always provide a field record layout. Don't assume field name and content are obvious. When the wrong fields are used in a matchback program, you risk paying for an extra round of processing if you discovered that your post-matchback sales are less than your pre-matchback sales.
Before starting the matchback process, do have your matchback vendor run both your mail tapes and the order file through standard address correction software. Standardizing addresses in both files will significantly improve your match rate. That is, instead of making the program try to match 1234 Hazel Grove Lane and 1234 Hazel Gr Ln, both your mail tapes and your orders will have the same software-approved layout.
Don't forget to create instructions. We've found that because matchbacks can be extremely complex, it's best to create written instructions for your matchback vendor. These will provide both parties with clear guidelines and expectations. Start your instructions with basic information from the files discussed above (e.g., mail quantity, total sales). You'll add to these instructions as you go through the steps detailed below.
Do quick checks prior to matching back. Check that your matchback vendor has the correct mail counts, particularly if you're using a different vendor from your normal mail processing vendor. The vendor should provide a report of name quantity by mail drop from the mail tapes. Check it against your final merge/purge records for each mail drop to be sure the numbers match.
Do check that the total dollars and number of orders from the sales file that your matchback vendor uses matches your idea of total sales for that time period. If you run a gross sales report and it shows, say, $850,000, be sure the order export file also totals about $850,000. Don't sweat small differences, which can be caused by exchanges and returns. But your totals should be close enough to help ensure that the correct data are going into the process.
Do decide what kind of output you need. This might seem like it should be the last step, but you'll find that determining the format of what you expect post-matchback will help you define the criteria for the process, itself (see below). These reports should include sales and orders by mailed group and mail drop, as well as key performance indicators, such as average order value, response rates, percentage of breakeven or contribution, and sales per book mailed.
|Mail Drop||In-home Date||Matchback Date Range|
|Drop — No. 1||Sept. 1-3||Sept. 1-Oct. 2|
|Drop — No. 2||Oct. 3-5||Oct. 3-Oct. 31|
|Drop — No. 3||Nov. 1-3||Nov. 1-Dec. 4|
|Drop — No. 4||Dec. 5-7||Dec. 5-Dec. 31|
Don't forget to define your matchback logic. Typically, mailers will match back unknowns and Internet orders only because they assume all keycodes captured during order entry were captured accurately. But if you suspect you're having problems with accurate keycode capture in your contact center, consider matching all orders against the mail tapes. Compare the matched-back keycode with the captured keycode to assess the extent of the capture-rate problems.
Do decide how often and when to perform a matchback. How often you choose to do a matchback will depend on your mailing frequency, budget and planning cycle. Many large direct marketers, particularly those with retail channels, do matchbacks as frequently as once a month. But for smaller mailers, I recommend you do a matchback at least once a mailing season, for example, spring/summer, fall/holiday or whatever seasons fit your mail plan.
Don't forget to define a date range. Once you've established your matchback frequency, provide your data processing vendor with a list of contact pieces (e.g., catalog, postcard) sent during the matchback period and the specific date ranges each piece covers. See sample chart (above). The matchback date range for Drop No. 1 is Sept. 1 (the start of Drop No. 1 in-home) through Oct. 2 (the day before the subsequent catalog in-home).
Do determine your match priority. This can be the trickiest part of the matchback process. By match priority I mean the rules you create to allocate orders and in what order you choose to apply those rules. The specifics of your match priority will depend on the kinds of decisions you'll be making for future campaigns.
For example, let's say Jane Smith ordered in December. And in December you mailed her both a regular catalog and a special VIP mailing. If you allocate her order to the regular catalog, will your boss cut funding for your VIP mailing? If you allocate her order to the VIP mailing, will the deflated catalog results imply you should cut circulation? This isn't just an office politics issue; it's a question of maximizing return on investment. The key to match priority is creating rules that enable you to maximize decision-making. You may need to experiment with your rules to discern what works best for you.
Do carefully review your preliminary output. If it's your first matchback, have your data processing vendor send output file dumps before running the entire matchback. A dump is a handful of sample records from the output file that enable you to verify that you're getting back what you expected. It also gives you the opportunity to make changes to your matchback logic if needed.
As you do more matchbacks, you'll get better at defining criteria and gleaning useful data from the results. But for now, don't get hung up on details. You'll be forced to make compromises, but the information you'll find will be useful.
So until the day that all customers beg to give us their keycode information, happy matching.
Tools and Technologies to Try
Accurately determining what level of matchback your company needs can depend on several factors: available resources, the specifics of your contact strategy and time constraints imposed by future planning cycles, to name a few. Following are three steps that can help you select a strategy and vendors.
Step 1: Identify marketing channels you'd like to include in your matchback. You get sales from several channels. Which channel's orders should you include in your matchback, and which should you omit? You probably get orders from direct mail (e.g., catalogs, postcards, flyers, special mailings), Internet, e-mail marketing campaigns, paid search engines and affiliate marketing programs. Choose the mix that's right for you based on the following information.
Step 2: Determine how much you're willing to invest. Advanced matchback tools can come with prices that may be a bit too steep for smaller companies and may not be necessary for your level of business. Determine at what point the cost of a matchback will outweigh the impact of the sales decision. Take the following into consideration:
- the percent of unknowns to your overall sales,
- the proportion of your contact strategy that includes outside list sources, and
- if you suspect your allocation of marketing dollars will vary significantly based on the information.
Step 3: Study and compare resources. The discussion below addresses several possible solutions.
Low-cost matchbacks. If your marketing channels are limited or your budget tight, look for low-cost alternatives to the standard technologies available. You'll find that most data processing vendors offer some kind of matchback capabilities. Talk with your current data processing vendor to determine what kinds of matchback services it offers. It already handles your mailing files and is competent (or you wouldn't use its services). Costs can range from $500 to $5,000 depending on the quantity of names mailed and your vendor's pricing.
No-cost, faux matchback test. If your budget is extremely tight, a good way to keep matchback costs low is to go through the exercise of allocating unknowns based on several methods such as percent of total sales across the season, percent of total monthly sales per group or percent of marketing budget spent. Compare the results of these allocation methods with your matchback results. If the results for faux and actual matchbacks are close, you may be able to do fewer matchbacks in a given time period by using whatever allocation method seemed to give similar results. Don't skip doing actual matchbacks too frequently, however, as buying behaviors tend to change rapidly.
Stand-alone products. With matchbacks becoming ever more popular and marketing channels becoming more diverse, several matchback solutions have become available during the past several years. One such solution, ChannelView by Abacus, provides you with timely access (as often as daily if you need it) to Web-based metrics that track performance for up to six channels, and by campaign or promotion. It even provides metrics on promotional investments by campaign. Costs can range from $10,000 to more than $30,000, depending on order activity and mailing volume. Abacus recommends its solution for direct marketers that mail six or more times per year; have annual circulation of 2 million or more; and have unsourced orders exceeding 15 percent. Visit: www.abacus-us.com.
BenchMark by Catalog Marketing Services incorporates SKU-level matching into its response-analysis program. This has proven a good solution for marketers who mail frequently and whose products vary between mailings. For example, what if a customer ordered in January, but ordered a product available only in the December catalog? A standard matchback process would attribute that order to the January mailing (assuming this customer received both mailings). By serving as a central database for both mailing- and product-level data, BenchMark can attribute the order to the December mailing. Visit: www.cmscms.com.
Matchback tools also can be added to your existing order management software. These programs offer immediate matchback capabilities by integrating mailing data and order information into a single database. When orders are received, they're automatically matched to a mailed record. Two examples of these software solutions are the Media Matchback Module, which runs in Response Direct Commerce Software by CoLinear Systems, and the Matchback Response Analysis feature of WiseGuys by Desktop Marketing Solutions. The latter links to several standard order management software programs. Costs for these types of tools can range from $1,000 to $3,500 depending on what commerce software you use. Visit www.colinear.com or www.desktopmarketinginc.com.
Other worthwhile resources include: Channel Match by Experian (www.experian.com/products/channel_match.html) and the various solutions and services available from CognitiveDATA (www.cognitivedata.com).
Remember that matchbacks are not an exact science: There always will be a portion of your sales that come from unidentifiable sources. But keeping that proportion of your sales to a minimum can offer significant benefits to your overall marketing strategy.
Terrell Sellix is vice president of marketing at McIntyre Direct, a Portland, Ore.-based, full-service catalog agency. Sellix combines her love of data analysis with a sensitivity to the human side of marketing that stems from her four years of marketing good health concepts in West Africa. For questions, contact her at: (503) 286-1400.