DALE HAGEMEYER: Yet, there are some companies with great stories to tell about how they have used statistical modeling to change the way they go to market. As they’ll tell you, starting with something statistically generated and tweaking it slightly, modifying it, offers the ability to help them focus more on outcomes –– the outcome being a scenario that generates credibility. They are able to say to the customer: “What if we do this? Is that going to work for you?” And maybe he’ll say, “Let’s try it.”
Now, obviously, in such a case you’re assuming risk –- risk that you take the space, spend the money, run the promotion, create the display –– and it’s not successful.
But I do think that, over time, we have to become a more outcome-focused industry. As opposed to just activities.
ALEX RING: I agree. The caveat is: It has to work. For a company to rely on lift modeling and optimization, to use it to actually drive its decision-making process, it has to work most of the time. If half the time I get a reasonable number, and the other half of the time I get a bogus number, I will not be as confident of the capability as I would be if 80 or 90 percent of the time the model suggests a good starting point.
And if, even once in a while, I get a number that is simply absurd, how much can I really trust the model? What if, for example, you go to a certain website to book a flight –- and half the time it doesn’t get you the lowest fare? Or sometimes you can’t get on the site? How long would it be before you stopped using the site, and went somewhere else?
In short, for this capability to be taken seriously, at least directionally it has to be mostly accurate.
DALE HAGEMEYER: I’d like to go back for a moment to the change management challenge, and suggest that you do some “ WIIFM”-based planning: Think about “What’s In It For Me?” in the context not just of your direct sales team, but of your broker. After all, the vast majority of CPG products are sold through brokers. At one time, it was my job to audit brokers, and I found that they spent 20 to 25 percent of their time just trying to figure out what their commissions would be. They’d be tracking orders as they came in, putting them into a spreadsheet, multiplying by
.02, or whatever the commission rate was, in order to figure out what they’re going to get.
Predictive modeling could appeal not only to brokers, but also to incentive-driven salespeople, because they could actually model their commissions.
And if you get them doing that, you may actually get them to do post-event analysis. Because let’s face it: How many brokers are really doing post-event analysis?
ALEX RING: I agree – IF the tool is only one of several that you expect the broker to use. Because you also expect to use their insight and their experience with the customer in the market. If we are saying: “Broker or salesperson, here’s the number; this is the answer,” what you are taking out of the equation is all of the “soft” information that a good customer relationship generates. In that case, you’re not getting full value on the commissions you’re paying.
DALE HAGEMEYER: Right. If it’s only about “The Number,” you might as well have “Chimpanzee & Company” for a broker, because all you need is someone who can execute against it.
I’m not advocating that. Again, the underlying premise is that the best people out there are making decisions, whether it’s the team at Wal-Mart, at Target, or the individual who’s a one-man or one-woman show at Nob Hill in California. These people have good intuition about the customer and the market. The idea is to give them tools such that the underperformers or average performers can essentially begin to match some of the best performers’ abilities –– conceptual, quantitative, intuitive. Especially people who are new, people who haven’t been in a certain market or people who haven’t worked with a particular account.
References:
Archives