In today’s economy, every CMO, marketing manager and director is being questioned on the value of each marketing program. The days of free-flowing campaigns with non-defined goals are long gone. Today, it’s all about running highly focused campaigns with direct, measurable results. To that end, the best way to ensure maximum ROI is to have a clear definition of your target audience. Marketing campaigns built on effective customer/prospect targeting have a much better chance of delivering a significant ROI.
That is why I asked Fred Pratt, Vice President of Marketing Analytics & Technology at AIR Marketing, to share some of his insights on what makes targeting an effective marketing tool.
MY QUESTION: How do you determine the parameters to effectively target prospects?
FRED’S ANSWER: It’s always best to clearly define the objective for the initiative: Is it customer acquisition, retention or growth (up-sell/cross-sell)? Once this is defined, it tells you the type of data that can be used. For example, acquisition targets cannot be based on transactional data, so you must rely on demographics or other data enhancements. The strongest attributes that any company can use is their own data. These will be most predictive and should definitely be used for retention and growth targeting.
Also, in selecting target criteria, it is important to exclude attributes that are directly correlated to the behavior being targeted. For example, if a resort is trying to target past guests for an upcoming summer stay, they should not use ‘months since last stay’ as an attribute because guests that visited last summer would, by definition, have a defined number of months since their last visit.
Q: How do you know when your target parameters are working?
A: You can measure the strength of target criteria in advance by using adequate sample sizes and predictive models. The models will only bring in target variables that are statistically significant. Once the model is built, it should be tested against a separated random sample to ensure the strength of the target. When applied to a campaign, it is always recommended to set up a control group to measure the real world performance of the campaign.
Q: When you develop a customer profile, how much information do you include?
A: That’s a difficult question to answer because it depends on the type of attributes and the number of possible outcomes within each attribute. Generally, a sample size of 10,000 or more will enable you to develop a solid customer profile, although a sample of 5,000 can be at least directional. The actual number of criteria used in a profile is usually from 7 - 12. Keep in mind that these are the result of sifting through sometimes hundreds of possible variables. Predictive models can sometimes bring in up to 20 variables; however, the most robust models have just a few variables.
Q: Is this dynamic? Meaning, do targeting definitions change with time?
A: Profiles do change over time, and as a result, should be updated at least annually. Some types of profiles are much more dynamic and require quarterly or monthly refreshes. An example is the profile of online resources or search keywords that a prospect uses when making purchase decisions. Some profiles actually get stronger over time, which can be limiting. This can happen when all marketing initiatives are based on the profile; then the only conversions fit the profile, thus it gets stronger and stronger. In this case, companies need to watch performance results carefully and recognize that trends are changing if campaign performance begins to fall off. This can be avoided by constantly testing and having multiple tiers of profiles.
Q: How dramatically does targeting improve marketing campaigns?
A: This can vary dramatically. Depending on the data and type of product/service profiled, campaign results will generally show a 40% - 60% lift compared to a non-targeted list. As with anything, there are tradeoffs. If you have limited marketing funds, a product/service that appeals to a specific type of customer and a large potential prospect market, then using predictive models can yield 100%+ lift. Targeting produces the best performance for companies with finite marketing budgets (almost everyone) and products/services that are not commodities.
Tags: Marketing, roi, targeting audience
Leave a Reply