When Smaller Mailings Mean Better Business
Cobb, Robin, Marketing
First there was profiling, then there was screening. Now there's modelling ... Robin Cobb reports on a mailing list innovation which is splitting the industry.
Profiling of mailing lists to give better targeting is well established. Now this is increasingly being coupled with screening techniques to eliminate individuals who, despite meeting other criteria, are unlikely to respond. It is a more sophisticated approach, designed to reduce waste -- but it is causing tension between list managers and brokers on one hand, and list users on the other.
How does it work? Take a lawnmower offer, for example. Profiling will sift the householders from the flat-dwellers. It may ensure that they have gardens. Perhaps, even, that they have previously purchased gardening equipment through the mail.
Screening takes over from there to eliminate those on the list who nevertheless have undesirable characteristics, such as not staying long at one address, or being broke.
But it goes further than that. Bespoke models can be built, arrived at through testing and experience from previous mailings, which place values on a whole range of criteria -- lifestyle, geodemographic, or both. These may include some absolute criteria which, irrespective of the rest, can result in automatic selection or rejection.
From all this, a judgment can be reached on cutting the resultant list from the bottom up, balancing the likely loss of responses against cost-savings in print, production and postage.
One agency which preaches the gospel of modelling is Colleagues Direct Marketing in Bath. List director John McGlone admits that, taken to extreme, it could result in a list of just three individuals who are certain to buy. But he contends: "Even on a good list, there are a lot of people who don't respond. Our industry spends a lot of time looking for the good stuff and not enough in dumping the dross."
While he acknowledges the contributions made by the proprietary lifestyle and geodemographic profiling systems, he insists: "The best way to build a model is to use factual data from previous mailings and tests you have done with that specific client. Then you can overlay the others."
For a model for the World Wildlife Fund (WWF), the starting point was lists of 2.4 million ostensibly appropriate names. The modelling reduced these to 1.4 million. An uplift of 14% was needed from the mailing to cover costs, including modelling and computer processing expenses.
This was substantially exceeded. Now the same model, with some further tweaking, will be applied to another WWF mailing.
McGlone emphasises that such detailed modelling is not appropriate for all campaigns. The set-up costs are high and may be prohibitive for small volume mailings. …