By Mary Conway
All names are not the same. That’s the basic truth underlying successful customer acquisition campaigns. Yet many marketers overlook this fact when they embark on a telemarketing program. They accept a shockingly low response rate from a customer or a prospect list as a “necessary evil,” instead focusing their attention on developing the most creative promotional concept and/or presenting an offer to the consumer via the vendor with the lowest hourly cost.
The result in far too many instances is hours of nonproductive calls to consumers who have little likelihood of responding to the offer. This wasteful practice can be greatly improved by tapping the growing power and sophistication of database modeling protocols, now increasingly being deployed in advanced teleservices programs.
Effective database modeling is built on the premise that all names are not the same – that better performing ones can be reliably predicted through careful list analysis, modeling, and prioritization.
It’s said that past performance is not a guarantee of future results, but it’s one of the most effective predictors of a successful telemarketing effort. Surprisingly, most teleservices programs overlook the prior performance of names on a list, failing to glean invaluable insights about how those leads are likely to respond in the future.
Analysis of a list’s prior response rates reveals important insights by answering questions like these: Does one geographic area, such as the northeast, respond better than another region, such as the southwest? Do urban markets outperform suburban ones? Do consumers residing in single-family dwellings respond better than consumers living in multifamily dwellings? Do credit scores correlate with response rates?
The analysis doesn’t stop there. Here’s another factor to consider: How recently has a name been called? Over-contact is a red flag in telemarketing. If a properly coded list indicates a household has been called recently – say within the last month – that name should probably be suppressed, as the likelihood of a positive response is low.
Many companies devise their own coding schemes for names on their customer lists. These too can be analyzed for varying response rates – even without knowing exactly what they mean!
Database modeling programs consider all these inputs and then apply a scoring mechanism that ranks expected response rates from highest to lowest. This evaluation is invaluable in guiding the hourly efforts of telemarketing teams – focusing them on fully penetrating prospect names with the best potential for sales success. Importantly, the assessment process must be continuous. For optimal results, as waves of a campaign are completed, the modeling program must evaluate performance on an ongoing basis, re-ranking remaining names before they are called.
This database modeling was recently employed for a credit card issuer. The client supplied a list of customer names, which it had ranked for expected propensity to enroll in a value-added program. This ranking was intended to drive the campaign’s list penetration strategy: over-penetrate the best leads and under-penetrate the worst ones.
Unfortunately, as the campaign unfolded, there were frequent instances where the performance scores did not correlate with actual results. Consequently, a significant degree of sales opportunity was not being realized. The company clearly needed a better database modeling strategy.
By overlaying additional modeling criteria on the list, a revised segmentation/scoring scheme could be generated. As the table below illustrates, the new campaign achieved significantly higher conversion rates, resulting in a cost-per-enrollment that was $10 lower than the previous strategy – a savings of over 22 percent.
|Sales per Hour:
|Cost per Enrollment:
As businesses relentlessly seek the best return-on-investment for their marketing investments, database modeling — skillfully applied – offers an impressive solution.
Mary Conway is chief marketing officer for DialAmerica.
[From Connection Magazine – June 2008]