This chapter is from the book
This chapter is from the book
This chapter is from the book
1.6 Value of Information
There exists almost an infinite number of metrics that could be calculated. Even the most quantitative marketer will recognize that more calculations don’t always help make better decisions. Thus one question a marketer may want to start with is: “when is a metric useful?”
A classic distinction is between data versus information versus knowledge. Data is what we have a profusion of in the world of big data. Data is in raw form and doesn’t tell us anything without being manipulated in some way. Information is data that has been converted into something that can be used by a human reader. Ideally, information gets converted into knowledge when a user understands and internalizes the information. Thus one way of thinking about the value of information is whether it creates knowledge or not. Data that is simply being stored is not currently valuable, but often has the potential to be valuable if approached in the right way. How can we extract the information from the data we have? (Clearly the marketer should ensure they have the legal and ethical right to use the data in this manner. Consent is usually a key consideration, but discussing law and ethics is beyond the scope of our book. The Direct Marketing Association has resources to aid such thinking, http://thedma.org/.)
One way to increase the value of information is thus to make it easier for users to convert it to knowledge. To do this we recommend considering how the information you have extracted, such as the metrics you have calculated, can be presented in a user-friendly way. There are now many companies, e.g., Tableau software, www.tableau.com, that specialize in translating information into visual representations. Such visual depictions are an excellent aid to allowing the user to more easily extract the message from the information you provide them.
An alternative way of thinking about the value of information is whether the information helps us take an action. Information is valuable only if it allows us to make a better decision. To cast this in terms of metrics, a metric’s value arises from its ability to improve our decisions in some way. Note this is a very pragmatic approach as the value of the metric depends upon what the user can do with the result. A CMO might find estimates of the value of the brand she controls invaluable when arguing for increasing the marketing budget with her C-suite colleagues. A more junior marketer, however, may feel that he can’t impact brand value in any significant way so knowing this number is of no value to him. The more junior marketer can, however, impact whether the product is on the retailers’ shelves and so may find distribution measures invaluable.
A related point is that people sometimes equate the value of information with the range of possible alternatives that the metric can take. Knowing the precise number for a metric that swings wildly can be very informative and thus valuable. If the metric never changes significantly, knowing its precise reading at any given point is unlikely to be very valuable. For example, information on the sales of a fashion item where consumer reaction is unpredictable can be exceptionally valuable for stock planning. Estimates for items with more predictable sales, e.g., safety matches, are less valuable because knowing the precise sales number is less likely to change the inventory order you would make without the more refined sales estimate. For items with very stable sales, your estimate based upon last year is likely to be good regardless of whether you calculate the precise metric for this year.
Testing is a critical component of marketing plans, but where should you spend your testing budget? What gives you the most information for your money? Scott Armstrong notes that this depends upon what you are trying to achieve. Sometimes you will want to emulate much academic research and drill down into a very specific topic. This can lead to very consistent estimates, also known as being “reliable.” This means every time you measure you get a similar result because you measure exactly the same thing each time you measure. In everyday life the electronic scale that weighs you every morning is reliable, you generally get the same result if nothing changes. This approach makes sense if it is critical for you to be very precise, if small changes in a metric would radically alter your plans.
More often, however, you aren’t sure you are measuring the right thing. You want to know how the firm is performing generally but you have a less than perfect understanding of what performance means exactly. You might be interested in your general health rather than your precise weight. Your weight is likely to be connected to your general health but is far from the complete picture. In such situations you are interested in whether the measures you are using are valid, whether the measures accurately capture what you want them to capture. To assess validity you are likely to want multiple measures, in which case you’ll spread your testing budget across a wider range of tests and are more tolerant of conflicting results. To assess your health you might look at your weight, your blood pressure, your blood sugar, the ease of your breathing, etc. These will sometimes point in different directions but put together they give a more comprehensive picture than fixating upon a single metric, however reliably the single metric can be measured.
To have valid estimates of hard-to-define concepts, such as performance, we often recommend a variety of tests and the use of multiple metrics. As we will discuss in Chapters 12 and 13 it is often possible to have one metric look very good while the true value of the company is destroyed. Testing multiple different areas and assessing different metrics may give you a less precise picture (it is less reliable) but is much less likely to miss a major problem (it is more valid).