What’s New with Match-Backs

Match-backs have become routine for catalogers today. This is the process where your order file is “matched-back” against your recent mail tapes in order to give credit to the proper source code. This will tell you where the business is coming from and which key codes should be given credit for the sale, even web orders. With the amount of business going to the web today it is next to impossible to track results to a specific source code without doing a match-back.

How a Match-Back Is Done

This process matches back orders to mailings using merge/purge logic. It allocates unknown orders back to mailed records based upon customer provided source code, customer number, merge/purge results, catalog in-home date and order date. The summary report includes key code, total mailed, total orders, total dollars, average order and dollars per book. Typical data needed to do a match-back includes the following:

  1. Mail files from appropriate time frame.
  2. A listing of all valid source or key codes. Examples: print mailings, e-mail campaigns, affiliate marketing, bounce backs, catalog request (inquiry) mailings, and other.
  3. Order header records for the appropriate time frame. Desired fields include name/address, customer number, source code, order date and order amount.

There are an increasing number of records which cannot be matched to a mail file. This is the result of increases in search and affiliate web marketing. These names tend not to be good catalog buyers. Therefore, catalogers are reducing the number of catalogs they mail to these customers. This puts a premium on match-back programs which not only provide response reports but are also able to identify which customers are responding to a catalog mailing and which ones are coming from non-mailed sources or offers. Those responding to a catalog mailing are then put on a contact strategy appropriate for catalog responders while the non-matches are sent at the most two more catalogs. In order to accomplish this, the information would have to be brought back into the file so that when we “pull” customer records for a mailing, we can tell the difference between web buyers and web buyers driven by a catalog mailing.

Why Do a Match-Back

Today with 40%, on average, of the business coming through the Internet coupled with the same 15% to 20% non-traceable factor we have been accustomed to for years, catalogers are lucky to trace 50% of the business to a specific source code. There are often lists that are made up of heavier web buyers and therefore, they may look like they do not work well, when they are actually profitable.  In this case, without a match-back you would never know this and any testing would have been without reason.  You might think you have trouble finding lists that work, when in actuality you have some winners to add to your continuations. In addition, you might have lists that look like they are falling off, or your total rentals look like they are trending downward.  This might just be a result of heavier web sales.  It sounds simple, but without truly knowing the level of performance of all segments mailed, you could be lead to some false conclusions that influence your marketing strategy.  A match-back will keep you on track and give you the level of confidence in the results needed to make sound circulation judgments.

How Often Should a Match-Back Be Done

How frequent you should run a match-back really depends on how often you mail and how you are going to apply the results and/or data to your marketing/circulation strategy. We feel that a match-back should be done two, three or four times a year depending on your degree of seasonality. For example, if your business if based on a fall/holiday season followed up with a spring season, doing a match-back twice a year is probably adequate. Some firms offer the ability to match-back weekly which we feel is overkill. It is convenient to have the ability to match-back often but probably not cost justified. Doing match-backs with the same service bureau who is doing your merge/purge work is logical since they already have your mail tapes.

Typical Results From a Match-Back

Prior to the Internet, the non-traceable factor ranged from 15% to 20%.  Then, we could simply allocate the unattributed orders and sales back across all source codes on a proportional basis. This will not work today because the allocation is far from proportional. Match-backs have shown us that it is not appropriate to give equal weight to the housefile, inquiries, coops, and rented lists today. In our example below, the match-back attributed 35% of web and unknown demand back to a specific mailing. By doing a match-back, we were able to identify $1,075,769 of gross demand revenue back to a specific source code. What’s more, 62% of the total was allocated back to the housefile. This reinforces the fact that the catalog is the largest driver of traffic to the Web and most of the business going to the web is really from your own customers. Approximately 17% of the non-traceable amount was matched back to a cooperative database and the same for the outside rental lists in our actual example here.

LETT Direct, Inc.
Date: 09-01-06
Housefile $1,598,219 $2,263,755 $665,536 41.64% 61.87%
Inquiries $77,422 $123,351 $45,929 59.32% 4.27%
Cooperative Databases $724,747 $907,297 $182,550 25.19% 16.97%
Rental Lists $672,787 $854,541 $181,754 27.02% 16.90%
Total $3,073,175 $4,148,944 $1,075,769 35.01% 100.00%

Keep in mind the affect re-mails can have on your match-back results. Let’s say you print for three catalog drops all at once changing only the front cover. For example, you might mail a Christmas 1 catalog in early October with Christmas 2 and 3 re-mails three or four weeks later. The match-back results will be allocated to the most recent drop which is Christmas 3 in our example. If your business is highly seasonal, this will skew results to the last mailing. Therefore, the results of Christmas 1, 2 and 3 should all be added together in order to spread the results more accurately across all of these drops. When it finds a match, the program will match to the most recent mail date. If it does not match that mail date, it goes back to an earlier drop, etc. We often do overlap our match-backs processing. For example, if we are doing quarterly match-backs and the most recent mailings is only 30% done and the one prior is 60% done, we still include them in the run. But, we also include these drops in the next match-back when they are 100% done in order to gain accuracy. Let’s say that you are doing a match-back for Spring & Summer and at the time, the % done were as follows:

Spring 1 = 100% done
Spring 2 = 100% done
Summer 1 = 60% done
Summer 2 = 30% done

You would want to include the summer drops in the current match back.  But, since they are not 100% done, you would want to also include them in your next match-back when they are 100% done. There could be quite a difference between what the results looked like by list at 30%, for example, vs. 100% and it confirms the total expected sales.

Cost To Do a Match-Back

The cost to do a match-back is in the range of $2.50 per thousand. The base price often decreases if the match-back is done as part of a housefile update. This number of books circulated can also affect the base price you are quoted from your service bureau. Subscription prices which provide for the ability to match-back on-line weekly range from $2,000 to $4,000 and up. This option is more expensive and more time consuming to manage for the mailer.

Match-backs are common place today. Match-backs do not replace the conventional source coding method of tracking results. Continue to assign and capture results by source codes just as you have always done and keep the pressure on your call center to ask for the code on the back of the catalog in the colored box. Match-backs are not perfect and certain assumptions have to be made which can affect the accuracy of your results, for example catalog re-mails as mentioned earlier. Match-backs are a good tool to help you better plan your marketing and circulation strategy.