In recent years, the number of sampling methods and programs available to brand managers have increased dramatically, and at the same time have become more sophisticated in their approach to sample delivery. Many of the new programs being introduced use modern marketing information systems to identify prime consumer groups and then a great deal of creativity is used to get the samples in their hands with a high degree of accuracy. Furthermore, distribution levels for many of the new programs being introduced have quickly moved from the tens of thousands into the millions.

For brand managers, the stakes are higher and the choices are greater than ever before. The significant cost of participating in sampling programs has resulted in the outright demand for quantitative research to substantiate the benefits derived through these types of promotional events. Many brand management business units are now requiring quantitative research to substantiate the success of a particular sampling program. In some cases, they are even requiring small "test" distributions (with research measurement) in the first year of participation before committing to the full distribution program in year two.

Brand managers who are committed to direct-to-consumer sampling have to make a lot of decisions regarding their product sampling efforts. But, the calculated or expected "payout" of a sampling program is what ultimately decides which programs are tried and then which ones are used again.

This brief summary will provide you with the basic factors you should consider when committing to a quantitative measurement of your direct-to-consumer sampling program.

Start thinking about the need for a marketing research measurement at the same time you consider participating in any specific direct-to-consumer sampling program.

In most cases, the executional elements of a particular sampling program will dictate the type of research study that can be implemented. So, assemble as much detailed information about the sampling program as possible. Confirm the method of distribution, the targeted consumer base, the intended location(s) of sample recipients and the timetable for the event (start/finish). With this information in hand, you can ask research professionals to design a measurement program that will be both accurate and cost effective.

Come to agreement on the objectives your company/brand group wishes to accomplish by participating in the sampling program of choice. Depending on the product being sampled, research objectives have been found to be quite different. But overall, several basic objectives tend to emerge. Not surprisingly brand managers who initiate sampling programs for their brands want to know....

1. The extent to which the program has reached the intended target audience;
2. The extent to which the samples were tried and who tried them; and,
3. The level to which the free samples inspired consumer purchase and repeat purchase of the brand.

This is the information that is commonly used to calculate the overall payout provided by the sampling program.

With this information in hand, you can then begin developing the measurement program that is best for you.

Before describing the types of research designs Burke applies to product sampling measurement programs, let's define a few terms:

• "Pre" Measure - A baseline measure that is taken prior to the sample distribution.
• "Post" Measure - This measure is completed after the samples have been distributed.

In most cases it is preferable to have both "pre" and "post" measurement data. The "pre" measure establishes the baseline to which all subsequent measures are compared. This type of sampling is very commonly used in general advertising tracking studies.

• "Independent"Samples - Two or more non-overlapping consumer groups.
• "Matched Group"Samples - The same individuals are interviewed in two or more measurement periods, thus forming a static panel of consumers whose brand purchase behavior is tracked over time.

The major difference between these two types of consumer samples is the level of precision that is inherently present in each sampling method.

"Matched Group" samples provide a greater level of measurement precision. Since repeated interviewing among the same individuals effectively reduces the error that might be expected in the research estimates, smaller "pre" to "post" differences are considered to be statistically significant. However, a great deal of care must be taken to ensure that the research questioning does not draw undue attention to the target brand, thus biasing the entire measurement process.

"Independent Group" sampling is probably the most popular method of sample construction, even though the level of precision in the research findings is always somewhat lower (given equal sample sizes) than that achieved through "matched group" sampling. Independent groups are appealing because there is absolutely no chance of the survey influencing consumer behavior. Furthermore, this method is typically less costly to complete and generally provides greater flexibility in the questionnaire design and interviewing process.

• "Test" Group - Consumers who actually receive the product samples.
• "Control" Group - Consumers who do not receive the product samples.

Obviously, every research project designed to measure the impact of a particular promotional event must include data collected among consumers who actually received the stimulus (i.e. the free sample).

In most situations it is also advisable to complete interviews among a similar group of consumers who did not receive the free sample. This is particularly important when there is a known seasonal effect that would logically be expected to have a positive effect on the key measurement criteria (e.g. usage and purchase of the particular brand).

When interviewing both "Test" and "Control" Groups in a measurement program, Burke works with the client to make sure that these two consumer groups are reasonably matched to one another. Also, when designing a research program that requires both "recipient" and "non-recipient" information, it is very important to ensure that these two consumer groups are reasonably similar to one another. Burke conducts an analysis of traditional consumer demographics (age/income/etc.) as a typical starting point for this comparison. It is also desirable to investigate overall category consumption behavior to ensure that neither group is disproportionate from the standpoint of purchasing specific brands in the target category or their frequency of purchasing brands in the category.

Combined with a specified data collection system, these six research options can be used to design research programs that fulfill the stated objectives of any sampling program measurement. Several of the more common designs used by Burke are outlined in the next section.

The first and most rigorous research design provides "pre" and "post" measures among consumers assigned to both "test" and "control" groups. The data collected in this research program is analyzed by comparing "pre" to "post" movements in both the "test" and "control" cells. If gains in the "test" group significantly exceed those for the "control" group, the sampling program is considered to be successful.

Since it takes into consideration any movements in the key evaluative measures that have occurred with and without the receipt of the product sample, this particular design is considered to be the gold standard for this type of measurement program. It can accommodate either "independent" or "matched group" consumer sampling.

However, since this research design requires four unique consumer cells, it can be very costly to administer. In some situations it is possible to reduce the overall cost of this research program by completing only one "baseline" measure to which both "post" measures are compared.

Another design that is commonly used includes a "post" measurement among both sample recipients and non-recipients. The statistical analysis of the differences noted between these two groups identifies gains that can reasonably be attributed to the sampling program.

This particular design can be recommended in situations where the consumer market is considered to be very homogeneous with respect to category and brand consumption. Furthermore, it's best to have both consumer groups selected from within the same market area. In doing so, other advertising and promotional expenditures for the brand can be assumed to have had an equal effect on the two groups. As mentioned earlier, consumer demographics and consumption behavior must always be compared to ensure that the two consumer groups are very similar to one another.

This design is commonly used when you have very controlled consumer panels, such as those selected from direct mail sampling programs. But, it can also be used as a backup design in situations where research is requested after the sampling event has already taken place.

The last design that is frequently used provides both "pre" and "post" measurements among sample recipients. This is a cost effective design that should be considered when no other promotion or advertising events are scheduled during the research measurement period. It can be used effectively to measure the success of certain types of distribution programs or in situations where cost savings is a major concern.

Before committing to a particular sampling program, it's a good idea to calculate the conversion level you will need to "break even" with respect to program expenditures and anticipated returns. At times this conversion estimate may provide you with a realistic "Go/No Go" decision criteria. If, based on market information, the estimated "break even conversion" level is unrealistically high, it would probably be a better idea to search for a more cost effective sampling program.

Burke combines survey research data with known category sales information to determine the "Break Even Conversion" level that must be achieved for a particular sampling program.

The "Break Even Conversion Rate" is calculated by dividing the Cost of Sampling by the Expected Profit that will be realized for each consumer that begins buying your brand as a result of the sample he or she received.

The "Cost of Sampling" for a particular program includes all of the out-of-pocket expenditures. These costs include the cost of product, shipping and handling, program costs and any other direct expenses that are incurred as a result of participating in the selected program.

Burke uses both research information and internal company data to effectively assist clients in assessing alternative sampling programs. Our "Payout" formula can be used to measure and then compare the relative success of a wide variety of different promotional vehicles, including: sampling, couponing, FSI's, direct mail and others.

With accurate consumer intelligence provided through Burke, marketers can effectively choose promotions that optimize the "Payout" for their brands.