How Do You Audit Your Service Bureau?

Service Bureaus play a most important role to catalog companies. However, dealing with a service bureau can be pretty scary. It involves a great deal of trust. Did they select the “right” names? Were a high percentage of the duplicates identified? How do you ever really know? There are always many challenges catalogers face when trying to audit their service bureau. Files sent from a catalog company to a service bureau are never “clean”. Often, there are duplicates within the file, NCOA changes are not applied back to the housefile, some records have invalid addresses, etc. One of the major reasons why catalogers use a service bureau is to identify and fix these issues and to get the file the proper condition for mailing purposes. Things can (and do) go wrong. We will look at ways to audit your service bureau so that potential problems can be identified and fixed before the mailing goes out.

Let’s say the size of your buyer file in July was 480,000 compared with a file size of 400,000 one year ago. Is this correct? Would you know if you were missing 10,000 or 15,000 buyers? There are many factors to consider when estimating your “new” file size making it difficult to identify small variances. That’s when we rely on a service bureau to run the necessary processing in order to provide us with accurate counts. If they tell us our new file size is within 5% of our estimate, we tend to accept this as correct. But is it?

So, how do you really know if you are missing buyers and that your segmentation, hygiene, selection and suppressions are all being processed correctly? Good questions! With auditing you can get to a level of comfort you feel good about. Shown below are some of the steps you can (and cannot) take to insure your housefile is being processed accurately and to your written (all instructions and specs must be in writing) specifications.

What You Can Do

  1. You should require your service bureau to give you record samples with counts by “drop” type, i.e., NCOA drops or invalid state/zip, foreign records, etc., and input/output counts for every step of the process. This should be done when and where it is possible for the service bureau to “drop” records for one reason or another during the file update. Typical steps to audit in this manner are as follows:
    • Input Counts – Do the input counts match the record count from the data you pulled and sent from your main frame?
    • Identify and edit the invalid records to be dropped from the file
    • Update duplicate rollups and/or drops
    • NCOA changes and/or drops
    • Suppression drops
    • Merge/duplicate drops
  2. You should pull sample records off your internal system and ask your service bureau to do a search in order to make certain the “right” customers are on the output file. At the same time, you’ll want to make sure there are not customer records on the output file that should not be there.
  3. You should look-up every customer in your system from your ink-jet samples to make sure each record falls into its proper recency, frequency and monetary value (R-F-M) cell.
  4. You also want to make certain that the do-not-mail records are not being mailed. This is not always easy because there will be exceptions where the service bureau has “rolled-up” transactional data on duplicates. This will cause the record have a higher R-F-M status than what you show on your system for that same record.
  5. You should match the R-F-M counts to last year’s R-F-M counts to make certain they are in the ballpark. If, for example, your “new to file” count increased by 20,000 over the prior year and your housefile response rates are about the same as the prior year, you can expect your 0-12 month file to increase by roughly the same 20,000 records. It is important to note that if you have a very high unknown rate it will be more difficult for you to estimate your “new to file” count. As a result, you might have to rely more on logic and estimates in order to verify counts.
  6. You should create decoys within your housefile. Enter a unique name (i.e., a fictitious name) as a buyer and upgrade that buyer’s status on occasion by showing repeat purchases for it. Such decoys can be checked against the service bureau’s file output to ensure its R-F-M matches the R-F-M you’ve created for the fictitious name. Monitoring the decoys will also help you know if and when the catalog was received.

What You Cannot Do

  1. You cannot read the logic and/or algorithms. i.e., the code, in your service bureau’s programming to make certain it is correct. You have to rely on them and on their programming staff to get it right.
  2. You cannot match your record count to the final output from your service bureau. This is due to the fact that your service bureau has run steps to roll-up duplicate records, update hygiene, suppress your do-not-mail file and more.
  3. You cannot run your own process to mirror the processing your service bureau does in order to validate the output. It would be nice if you can go through the same steps to get approximately the same end result but this is not realistic or practical.
  4. You cannot ever be 100% sure that you have the cleanest, most accurate file possible. But, you can come close. Service bureaus perform an important function and they do a great job giving you a good, clean file for mailing efficiencies.

Several of these audit steps require access to your mainframe. If you outsource your circulation to a third party, they can only do part of the audit. The rest needs to be done internally by your computer operations staff. Any steps which require validation to the mainframe will need to be done by someone within your own company.
The importance of getting your catalogs to the “right” people is critical to your success. By auditing your file and the merge/purge process, you will be able to help prevent buyers from slipping through the cracks. It is important to do everything possible to avoid potential problems which might not be so obvious until it is too late.