As direct marketers, we know the importance of well-designed testing on the critical factors of list, offer and creative. However, many of us run into problems when trying to test effectively in outbound telemarketing, as there are added variables when the call center and the reps are in the mix. Let’s take each category of testing separately.
The list. When planning list tests over the phone, the main concern is how they are named and managed in the system. Most predictive dialer systems can merge lists so they are called randomly. This makes the test transparent to the phone reps. If the system is set up properly, results can be reported by individual list.
If, for some reason, the lists need to be kept separate, it’s important to name the lists carefully. Reps tend to skew results based on their own bias. If they think one list will perform better, it will. In one instance, four lists were given letter designations randomly, starting with the letter A. The reps thought the “A” list was the best and the “D” list was the worst. Not surprisingly, that’s how the lists performed.
The offer. An important difference between mail and phone offer tests is that the phone rep sees both offers. The price test provides a great example. In one test, two price points were tested for a product, $29.95 and $34.95. In the mail, the $5 price increase had a 10 percent drop in response rate. In contrast, the phone sales rate dropped 35 percent. The only difference was that the reps knew that the $5 increase was for the same product. A second test was done where management met with the reps to explain the price increase in terms of the additional cost of production. In this test, the drop-off was much more in line with mail results.
Premium testing provides another good example. If phone reps are not used to offering a premium with the product, they often see it as an attempt to bribe the customer. The premium should be positioned as a bonus to get the customer to buy now, not as a substitute for selling the product itself.
Creative. Creative testing in outbound telemarketing is, for the most part, script testing. Unless your script is verbatim, this is probably the most challenging type of test to execute. Where reps have the flexibility to deliver the message in their own words, testing cannot be done verbatim. This not only skews the results but depresses the overall sales rate.
The best way to handle these tests is to decide on the key elements of each script test, create guidelines and train the reps on the guidelines as well as the “spirit” of the test. Monitoring should be increased to ensure that reps keep to the test instructions.
Testing methodology. There are several methods to manage tests, and all have pros and cons. One involves dividing the reps into two test groups and switching halfway through the test. The critical element here is to ensure that both groups have an even distribution of high, medium and low performers. A plan also must be in place for replacing reps that leave the company during the test. The best method is to add an equal number to each group to maintain the balance.
Another technique has all reps calling one test for a period of time, then switching to the other test. There are two drawbacks to this. First, a significant event may change the results from the first period to the second, such as a holiday or change in economic conditions. Second, the list penetration can skew results. When a list is first called, many wrong numbers are cleaned out. At the end of the list, there are fewer contacts and a much higher percentage of answering machines and no answers. Setting up two completely separate lists for the test can minimize this problem.
One final method involves alternating the reps between the tests frequently. This has the advantage of being the most random. However, it often yields skewed results because the reps never get comfortable with the test and can’t build momentum.
Results can vary significantly by day of the week and time of day. If one test has more calling on Friday when contacts tend to be low, this will skew the results.
Also, the reps’ monetary incentives can affect results. Reps participating in a test should not be penalized. Their incentives should be adjusted so that they make about the same amount that they make on production. Also, if one test is doing poorly compared with the other, reps tend to steal the words or the offer from the winning test. This can be minimized by offering group incentives for the combined results of both teams.
Last, ensure that the scope of the test is reasonable. Execute only one or two tests at a time. People have a limited capacity to absorb and perform different tasks at an acceptable level. If there are five test cells, then the reps need to expand beyond their usual single offer or script to handle five times the amount of input. Furthermore, test only one variable at a time. This should be obvious but requires expanding the definition of variable to include time of day, dates, multiple sites, phone reps and list penetration.
Managing tests in an outbound environment is challenging. It can be done effectively by adhering to set call parameters for each test cell and ensuring that the reps on the phones understand the concept and reason for the test.
