I have always been good at math. I remember the summer of my birthday, my parents went on vacation to the Bahamas leaving three of my siblings and me under the watchful eyes of my maternal grandparents. Being grandparents, buying gifts was their prerogative. While at the grocery store, my waist-level eyes caught a glimpse of something that I just needed to have. Needless to say, my grandmother was more than delighted to accommodate my desire, buying me my first set of multiplication flash cards. From that point in my life, other than the occasional dyslexic shift in digits, counting has been a skill at which I excelled.

So, it seems odd to me that when clients inquire about how to count, I fumble for a succinct answer. The counting inquiry: "How does one count response to a marketing program?" I usually surprise myself with an immediate reply of deafening silence to the inquirer while inside my head a jumble of words such as: gross, net, over-control, statistically significant, single and multiple attributions, direct and inferred. Ultimately, I suggest that we find 30 minutes, a room and a whiteboard so I can organize that jumble of words and share ideas with the inquirer.

Today, I thought it might be nice to write down several rules on how to count, so the next time I get that question, I can hand the inquirer a laminated copy of this column. With that in mind, let me try to organize a coherent reply to the question about how to count.

Rule #1: Define the Response Behavior(s)

The first consideration in counting is specifying what is being counted. For marketing programs, this means clearly defining what constitutes a response. In many cases it is the act of purchase. In other cases it may be an inquiry, and in situations where a marketing program is trying to engender loyalty, it may be the nonact of attrition.

Additionally, for more advanced response trackers, there are a series of potential response behaviors that are tracked. The series of behaviors typically correlates with the prospect/customer's buying cycle - awareness, consideration, interest, purchase and loyalty, for example. This multiple response behavior approach permits the marketer to understand which stage(s) of a buying cycle a particular marketing program can drive the prospect/customer.

Rule #2: Determine the WhoOn the surface, this rule seems like a no-brainer. One simply looks for instances of people who received the marketing message by responding. In many cases, it is that simple as would be illustrated by tracking the response to a pre-approved credit card invitation. In other situations, the marketing material may not have an individual identifier on it (or the response or fulfillment systems do not have a mechanism to capture it). "The who" must then be inferred - derived based on a set of rule (see rule number three). Typically, this means looking for name, address or household matches to identify the responders.

Additionally, in the age of viral marketing where we pass on emails to others who might also be interested or post some of the "great deals" on Web sites for the world to avail itself of, response may be impossible to clearly identify as it may look much the same as nonmarketed prospect/customer behavior. In this case, response attribution at the individual/household level may be impossible and the marketer should revert to using math to unravel the response mystery (see rule number four).

Rule #3: Match the Response Behavior to "The Who"Armed with the response behaviors to count and the list of folks marketed to, response behavior attribution is ready to happen. The simplest form of response attribution is often referred to as "direct." Using a direct response attribution mechanism, there is an identifier that is associated with the response behaviors tracked and the list of folks marketed that permits an associated between the two to be made. Again, a pre-approved credit card mailing is an easy-to-understand example.

When there is no identifier available, usually due to it missing from the response behavior data, response attribution is "inferred." Inferred responses are based on the application of a series of business rules. Typically, the business rules are used to determine the level of certainty that "the who" responding was part of the marketing list. Most often this involves using individual identifiers, household identifiers, addresses, phone numbers and email addresses to establish a probabilistic link. Above a cutoff probability, we say it is a match. Below, the cutoff we say it is not attributed to a marketing activity.

Additionally, some organizations who can identify that the response behavior could only have resulted from having been exposed to the marketing materials but for whom the probabilistic link cannot be established may choose to count the response behavior as a "viral" response.

The last issue when using inferred response attribution is deciding how many or which marketing program gets credit for a particular response. Often, organizations begin by allowing multiple response attribution as it is often the politically expeditious approach. However, the downside of this approach is that adding up the impact of marketing programs across the enterprise will result in an overstatement of marketing's impact unless multiple attributions are tracked and removed from these reports. World-class marketers either establish specific rules to define the best match or perform a partial attribution - crediting multiple programs with a factional response.

Rule #4: The Rest is Just MathFrom this point, the raw information has been established to let you answer any of a series of measurement questions using the appropriate metrics - counts and response rates. To understand the level of activities resulting from the marketing program, one need only add up the response behaviors of interest for the individuals contacted as part of a marketing program (i.e., gross response). To understand the impact of a marketing program, one need only add up the responses for the test group in comparison to its control group (net response or lift).

Life seems pretty simple by time one gets to rule number four. However, looking at my laminated list, I am wondering whether it was wise to let any of this out of my head at all. Perhaps the reply of deafening silence while doing multiplication tables in my head would be best for everyone. Happy counting.