Data Assumptions, Part 2: Taking Action

Last time, I examined the dangers and necessity of making assumptions as a professional marketer. Analytics data and other information will never be 100 percent accurate and complete, so eventually you must take a leap of faith. Fortunately, making that leap can be easier when you know how to get the most out from data you actually have.

As a quick recap, three big assumptions, which are part of a decision chain and can have a huge impact on your business, are:

Analytics data are accurate

Offline conversion rates are correct

A lead value is valid

For each assumption, there are ways to ensure valuable outcomes even with less-than-perfect data.

Keep in mind: be consistent when building models for behavior. That helps mitigate data that may be slightly askew. Take, as an example, a product family that requires a customer to leave the Web and enter a store to complete a purchase. This could be cars, boats, swimming pools, or tractors. For each, there’s a speculative leap that occurs when the customer reaches the online path’s end and heads to the store. The ability to predict the lead conversions and value with reasonable accuracy is sufficient to drive the business forward, or not.

What’s key is the ability to assign a value to each step along the consideration funnel. Rather than just assume an overall value of a lead — let’s say $130 — break down specific consideration points and assign a value to each. This might mean downloading a brochure is valued $20, watching an online demo video, $35, and talking to a sales rep is worth $75. Build your model with each value measured against each activity, because you’ve tagged each activity in your Web analytics software. This exercise helps you get finer gradations in decision making and avoid having a single big assumption cause the entire model to collapse.

What’s more, it’s essential to apply a degree of “fudge factor” to this process. Look at historical data, consider your gut instinct, and revise the final numbers. Take into account the reality of customers performing multiple activities along the consideration path as well as offline channel reporting inaccuracies. Your team and data can guide you here, but consider applying a 10 percent discount to the final numbers to avoid overstating results.

You’re always going to make assumptions. You’re always going to take action. Think about the entire business, both online and offline, and take as fine-grained approach as the data allows you to offset the danger of big assumptions. You don’t need to calculate actual dollars; you can create a token system and perform calculations that way. If someone downloading a brochure is worth twice as much to your business as someone doing a feature comparison, then factor that into the model.

Bottom line: when you act, you must always assume something. It’s better to assume you forgot to turn off the oven than to assume the house is already burning. One of those assumptions lets you take action now and prevent damage later. The other’s a disaster.