What did you have for breakfast today? In my house, we have a lot of choices for breakfast. There are at least 8 boxes of cereal, multiple flavors of pop tarts, and rice cakes. Not to mention, leftover orange rolls and a bag of mini donuts.
Why do we have so many choices? There aren't many people I know who want the same breakfast food every day – some days we want cereal, some days we want bagels, some days we want steak and eggs. We like the variety… and ultimately allocate our choices for breakfast across many different options.
This principle of allocation holds true for business decisions as well. How many businesses do you know that are using a single agency for all of their advertising? None! Instead, they typically have a basketful of agencies to use, and select the one agency that makes sense for a given business need. Their overall spending is allocated across the suppliers they approve.
But the principle of allocation does not only apply to professional services. Shippers will spread their packages across multiple providers depending on price, size/weight of package, access points and delivery time options. Eye doctors will prescribe different types and brands of contact lenses depending on patient need, timeframe of use and even cosmetic impact. Consumers will pick different brands of greeting cards depending on the creativity of the card, the occasion for the purchase, price and retail convenience (online and offline). In all of these cases, decision makers have multiple options to choose from, and allocate their spending across the options as needed.
So, why aren't we researchers doing a better job of incorporating allocations into our choice modeling efforts? The standard approach when using trade-off and conjoint-based tools is forcing respondents to "pick one" option. While there is a proven track record for this approach, it is not always how the purchase works in the "real world".
At KS&R, we have been using an allocation approach (within Hierarchical Bayes architecture) for many years now. In each case, we establish a framework for the choices within the competitive set. Respondents are then given multiple "test" situations where they are asked to allocate their purchases across the options offered. Each "test" situation provides a full category view – i.e., allocations across the choices will add to 100%. Ultimately, a share of preference model is created based on the allocations, and we are able to simulate category-level situations for business strategy and planning.
Our clients have found this approach to be highly intuitive and appropriate given their business objectives. Over time, we have used the principle of allocation to inform decisions about new product development, features evaluation, service bundles, and pricing/price elasticity measurement. It is not the perfect solution, but it does an effective job of mirroring purchase behaviors, which is especially important in categories where purchases are really an allocation of spending across providers, and not a single choice.
Speaking of choices, I had a low-fat strawberry pop tart today. Tomorrow, I'm going for the orange rolls.