Outside List Optimization: Does it Pay?

Finding the perfect conditions for fine-tuning lists can help define your response.

The process of optimizing outside rented lists, commonly called Marginal List Optimization, is known to increase response rates. Abacus claims the lift ranges from 10 percent to 15 percent, which is consistent with my own experience. However, there is a cost associated with optimization.

So can we say that the procedure is cost-justified? In other words, is the incremental increase in response and in the revenue per catalog mailed greater than the expense? In most cases, I feel the answer is no, which I will explain. There is, however, a place for list optimization when used properly.

Outside lists, which perform at or above the incremental break-even point, are normally not optimized. Optimizing these proven lists would only increase the breakeven point, and for little or no reason, since they are already performing at acceptable levels. That’s why most catalogers optimize the lists that are slightly below the break-even point, i.e., their “marginal” lists. It is also common to optimize compiled lists, such as those from Polk. Optimization identifies the catalog buyers who are on these compiled files and ranks them best to worst in terms of which groups are most likely to purchase.

A certain level of results will be achieved without optimizing. Let’s say, for example, that an average order size of $60 can generate a 1.5-percent response without the added step of optimization. With optimization, let’s assume the response rate increases to 1.68 percent, generating a 12-percent lift in response. Without optimization, $0.90 per catalog mailed is generated vs. $1.01 per book with optimization. In this example, optimization resulted in an $0.11 per catalog mailed increase. This is the incremental increase associated with the optimization process. This incremental increase needs to be enough to offset the cost associated with optimization. If not, optimization is not cost-justified.

How it Works

Here is what occurs. As an example, let’s optimize 10 lists of 10,000 names each for a grand total of 100,000 outside names. Optimizing will identify all of the single purchasers (one-time-only buyers) who will be put into a separate group. Within this group, a mailer may elect to mail them all under separate key codes or may choose to eliminate the bottom 10 percent altogether in order to lift response. Regardless, the cost for optimizing these names must be cost-justified, or it is not worth the added expense.

There are cases where optimization works; please refer to “The Economics of Outside List Optimization” chart. The area highlighted in yellow details the results needed to reach the incremental break-even point. On a mailing to 100,000 outside prospect names, a 1.82-percent response rate and an average order size of $60 ($1.09 per catalog) is needed to break even.

The first purple highlighted column details the results needed to achieve breakeven, if paying the full rate card for the names. As you can see, the response rate now needed is 2.95 percent with an average order size of $60 or $1.77 per book. If renting names, which in turn need to be optimized, a 62-percent increase in results is needed to pay for this expense. This compares with the 10-percent to 15-percent lift expected. Obviously, this scenario does not pay.

The second purple highlighted column assumes renting outside lists on a “net name” basis. This means only having to pay the list owner for the names actually mailed vs. the names rented, i.e., gross names ordered vs. net names mailed. With this scenario, only an 11-percent lift in response is needed—much more realistic. What is not realistic, however, is the expectation that list owners will rent on a “net name” basis. This is most unlikely, especially considering these, for the most part, are marginal lists for our offer and the quantity of names ordered would not warrant a net deal.

The Best Economic Scenario

The “Cross Member Model” column makes the most sense economically. Here, only a 10-percent lift in response is needed to pay for the investment. A cross member model is when another member of a co-op, such as Abacus Alliance, allows its list to be optimized on a “net name” basis. An order still needs to be placed through the list broker. The cataloger requests a cross member model not knowing if this firm is a member of Abacus. Since Abacus is a “blind” cooperative database, it neither publishes a listing of its members nor reveals this information. So, the broker sends the order to the owner’s list manager and on to Abacus if approved. While the economics of this scenario make sense, several of the lists wanted for optimization would not be members of the Abacus Alliance, including publishers of compiled lists, such as Polk. However, this is the option I prefer and the one that makes the most economic and practical sense. Abacus claims this is a stronger model, too. It is more predictive because of the large amount of transitional data that comes into play, since all the files reside at Abacus.

Optimization costs are based on the net output of names vs. the gross input. To optimize an outside rented list, a cataloger needs to rent 100,000 names from any given list. Abacus selects a minimum of 25,000 names to make the economics work. That’s because Abacus charges $40 per thousand net names outputted amounting to $1,000, which is its minimum charge. If a single list of 10,000 names is optimized, the same $1,000 minimum is charged as would be for 25,000 names outputted.

Let’s take a compiled list such as Polk’s. Optimization might work with a list such as this due to the fact that much less is paid per thousand names rented. What’s more, these are not proven mail order buyers, and therefore there is very little chance for success mailing them on their own. By modeling or optimizing these names, we can identify the mail order catalog buyers within the universe of names available to us, which will result in a lift. We need to work through the economics of this, but the application seems to make sense.

Taking into account the different types of modeling techniques, list optimization does not consider as many variables. In fact, this process only considers two or three key variables. Cross member models, on the other hand, use housefile modeling, which is the more advanced form of modeling done by Abacus. Since housefiles reside at Abacus, there is much more data available to Abacus for this process.

In summary, outside list optimization can be considered in the following examples:

• When lists are obtained on exchange from another list owner.

• If a “net name” arrangement can be negotiated.

• When compiled lists are used (providing the economics work).

• If a cross member model can be done.

• If we know we can achieve lift in excess of the cost differential needed to achieve breakeven.

Abacus, for example, developed its optimization model in order to identify single buyers post merge/purge. From my experience, this Abacus product works best on subscriber files and/or on compiled lists. It was designed for homogeneous lists. Trying to model several different lists is not the best use of this process.

I hope this provides a good understanding and explanation of outside list optimization. It is a good marketing technique to employ in certain cases, providing the economics work.

Stephen R. Lett is president of Lett Direct Inc., a catalog consulting firm specializing in circulation planning, forecasting and analysis. He is also on the faculty at Indiana University, where he teaches direct marketing at the MBA level. He can be reached at (317) 844-8228 or by e-mail at slett@lettdirect.com