We use cookies and other similar technologies (Cookies) to enhance your experience and to provide you with relevant content and ads. By using our website, you are agreeing to the use of Cookies. You can change your settings at any time. Cookie Policy.


Defining Justifiable Targets for Analytics Success Part of a Continuing Series on Straightforward Analytics

Originally published May 26, 2011

Our expectation is that the team focusing on analytics is doing so to help achieve some business objective, even if that objective is stated in high-level terms. Yet in order to determine whether the analytics technology is going to yield some benefit, as we have discussed in previous articles, the expectation of achieving the objective must make sense within the context of what is feasibly doable in the organization.

So although we are looking to define “targets for success” through the use of analytics, the operational word here is “justifiable.” I suggest this largely as away to combat two common scenarios. The first is what you might call “directive by fiat.” The second can be referred to as “technology overkill.” By looking at some (sanitized) examples, we’ll review how each of these scenarios reduces the effectiveness of your investment in analysis, perhaps even leading to some bonehead conclusions.

Directive by Fiat

Let’s start with an example of “directive by fiat.”  In this scenario, a key stakeholder asserts a target for success that has no realistic basis. As an example, consider a financial services company does some high-level analysis to measure both the total and average numbers of products or services that each customer has. The result is that the average number of products per customer is 3.5; as a result, a senior manager insists that this average must be increased by 10% so that the new target is to increase the average number of products/services per customer to 3.85.
Here we can start asking a bunch of questions:
  • What motivated the manager to assert that the average number of products be increased by 10%? (justification)

  • Can the market support increased product purchasing? (business environment)

  • Are there any qualitative differences between those customers with a high number of products and those with a low number? (consideration)

  • What products should be marketed and sold to increase the number of products each customer has? (approach)

  • Should the sales team focus on a small number of customers and sell them lots of products, or those with a single product and get them to buy just one more product? (method)
Without being able to answer all of these questions, the target does not make sense. Actually, without being able to answer any of these questions, the target does not make sense. Yet this happens all the time, whether it is an arbitrary determination of increased sales, determination of operational efficiencies, or increased productivity.

Technology Overkill

Our second scenario is expecting technology to tell you something that you could have reasoned by common sense, without expecting technology to provide the answer. I recall a presentation I heard a few years ago discussing the use of predictive analytics for sales productivity at an amusement park. Apparently the analysis demonstrated that there was a correlation between inclement weather (specifically, rain) and decreased point of sale productivity. A review of the business situation showed that when it rained, people moved out of the rain into the closest areas of shelter, which were mostly stores. Some were induced to buy more products, mostly as a result of being in these stores. Additional purchases meant that more people were standing in line to buy stuff, and that meant that the usual number of open registers was not sufficient to handle the extra load. Voila – decreased point of sale productivity.

But do you really need an analytics environment to tell you this? On the contrary, one might say that this is just common sense, as are many other conclusions “discovered” by analytics systems. And if we are looking at small and midsize businesses, the owners or executive managers are likely to have their fingers on the pulse of the organization to know the most significant operational opportunities.

Define Justifiable Targets

So to best understand what is meant by the third phase of our process preparation “define justifiable targets for success” (as discussed in the initial article on straightforward analytics), specifically for the use of analytics, let’s break it down into more discrete objectives:
  1. Consider the areas of the business where there is room for improvement but for which there are no truly obvious process improvements. Let’s use our common sense for effecting optimization first.

  2. Use the measures described in the second phase or process preparation as more than just a baseline, but rather as the starting point for understanding the limits of existing processes and methods. Seek to understand what influences the current state of the environment.

  3. Think about the constraints within the environment that exert downward pressure on success. Determine whether there are external realties that will prevent additional positive business value from being achieved.

  4. What alternatives are there for making change? Define approaches that can be used for comparison.

  5. Note the degree of effort necessary for improvement. If there is an uphill battle, will the effort be worth it?
As noted a few months back, the measure of success is not just a target value. Instead, it needs to be defined in terms of how it addresses the corporate need. By using our measures as more than just a thumbnail sketch of the organization’s practices, one can look for the best opportunities for setting success targets.

Recent articles by David Loshin

 

Comments

Want to post a comment? Login or become a member today!

Posted May 27, 2011 by shobhna bansal

Hi David, its a great article. However, I would suggest that it would be interesting to define the criteria for selecting the process for improvement. Analytics using pattern discovery and prediction techniques should support by giving us what are those influencing factors. Its not always known beforehand. Regarding point # 3 about External Realities, I believe its quite tough to include/builde into one' analytics model. Can you suggest how it can be done?

And my last point is that making change is part of business process improvement, the feedback which goes from analytics system to business process. BI is good at measuring the impact of the change.

Is this comment inappropriate? Click here to flag this comment.