Jeff Zabin is well known for his customer data capture and marketing techniques. Currently, Jeff is the Director of Marketing at Fair Isaac Corporation, a company whose analytics help brands acquire customers more efficiently. Jeff is also the co-author of two books: The Seven Steps to Nirvana, and Precision Marketing.
Zabin addressed the iMedia Brand Summit in Florida in February. Read the first part of that presentation here. Here's more:
Zabin: Can you tell I was a history major in college? I am going to talk a little bit about the history of the scientific method, because I think it is interesting. I won't give you a lesson about how history repeats itself. But, you know it all started, as I traced this back, I found that it all started with this Renaissance statesman and philosopher named Francis Bacon, and he came up with this notion of a hypothesis. (I am very familiar with this because I just did my son's Science Fair experiment last week. And, I am sorry, I did not bring the solar-powered backscratcher to show you.) But, he came up with this idea of a hypothesis. It is basically, you know, let's come up with an assumption about how something is going to happen. Let's create an experiment around it. We'll test it. And, then we will refine it and we will go around again and see what happens. So, eventually that evolved into what we call a Plan-Do-Check-Act cycle. It is a closed-loop process. It is in mathematics what you would call a recursive loop.
The first person to apply that in the context of business was W. Edwards Deming. He was a professor of Statistics at Northwestern University. He made a big splash, at least off the waters of the Pacific Rim. He introduced this methodology to Japanese automobile manufacturers. That could account for the reason that they were able to maintain superiority in terms of manufacturing up until recent years.
I really think that all we are doing now is applying this very same concept in the context of marketing. It is a scientific method all over again. It is about collecting data; about segmenting the data; executing; measuring effectiveness; and, I think that is key and that is what everyone seems to be driving toward today. How do we measure effectiveness? You know, Peter Drucker said, "If you can't measure it, you can't manage it." I think that is absolutely true. And, then refining it for the next go around.
And, I think there are really three enablers of what I call the precision marketing circle, or cycle. It is hardware, it is software and peopleware. Hardware is your basic infrastructure, your server farms, you know, all the hardware you need to house your customer data. Software is all the different tools and capabilities, in terms of, you know, your channel management tools, your campaign management tools, your analytic tools. You need a very robust customer data repository for housing all that information. It should have a flexible-style architecture that allows you to manipulate the data and query against it. And, then you need all kinds of recording tools -- preferably real-time recording tools -- so that you can track all of the output of your marketing programs. And, then this all needs to be integrated together using customer data integration technologies. And, you want to make sure that it breaks down all of your data silos again, so that you can maintain that view of the customer relationship.
Then, finally, there is the peopleware aspect and … we have 350 white-coated PhD statisticians and behavioral scientists who sit around all day and crank out equations and complex algorithms, and talk about fractional factorials, and God knows what they do in there … SpiderBots? I still don't know what a SpiderBot is.
So, I want to talk about ways of measuring. And, there are lots of ways to go about measuring, using precision marketing, using data-driven marketing programs. I think Kraft actually does a very nice job with its Food and Family magazine. Actually, here, this is an old slide. I have "2 million … sends it out to 2 million customers." Actually, it goes out to 12 million customers. This is when it was first launched, now it is a year later. Twelve million customers. I think there are something like 40 different versions of the magazine, including Comida y Familia, which is the Spanish version that goes out to 8 million families. And, they send these magazines out in all these different flavors. They use IRI scanner data. A week later they can tell, with a pretty good degree of precision, the impact of the magazine in different markets.
We have all seen the coupon redemption approach. And, that has been around ever since the invention of free-standing inserts and UPC codes. But, it is another good way. By using unique identifiers you can track the value of a marketing campaign.
I am going to go through a case study I think is very interesting, because more and more companies are trying to understand, "What is the value of our website?" It is one thing to say, "Yeah, we get lots of click-throughs. We get lots of hits. Lots of people are visiting our site. They seem like they like it. They are spending time there." But, does that really mean anything? I mean, at the end of the day, it is all about … everything should translate into dollars and cents. So, how do we really understand what the value is of that brand awareness? How do we really understand … you know, what is, how does that impact our sales volume and our revenues?
This was the very question that the consumer insights group at ConAgra was asking; and we worked with them very closely to create this design study. And, we also engaged PDI/Knowledge Networks on the retail side. (You know, they work with a handful of major grocery chains and they capture their loyalty card data.) And, so the idea was, let's try to capture the data on our website, people coming into the website, giving us information about themselves. And, then, let's match that to the data on the retailer's side. And, let's see. Let's really try to understand on a very granular level, to what extent the people who are using the website are actually going out there and increasing their consumption.
We did this cross-reference, and we had a hypothesis (and now you know what hypothesis is), going into this. And, this was just using information/benchmarks that were already out there. But, our understanding was that if you go onto a website, about 10 percent of those who visit a website are going to spend a little time in there, give us their name, and those are going to result in 9 percent incremental sales lift. Of those, a small fraction will continue to stay in there, they'll come back on an ongoing basis, and that is going to result in a 20 percent revenue increase. So, that was sort of our hypothesis going into this experiment.
So, we did the experiment, and what we found actually was that those benchmarks were lower than the outcomes of our experiment. We actually found that -- we did this across three different brands -- for the first brand the volume lift was actually 40 percent: twice what the benchmark told us. For Brand B it was 28 percent. And, then, for Brand C, I think we didn't have enough data. And, so we see the ROI was very high in that first brand. It was quite a bit lower in the second brand, but as you look at the results, you begin to realize that scale is very important. There are a lot of fixed costs when you first create this website and these mechanisms for capturing the data and for running the program. But, after a while, it is an economy of scale and it pays for itself. So, you need to get to that break even and then, you know, it is a free ride. So, that was the lesson we learned here.
And, then, ultimately we were able to … so, here is the final outcome. It was: a pre-study hypothesis -- 20.3 percent, and average was 22 percent for our studies. So, it was higher than the benchmark and well worth the effort. And it told us that we needed to continue going down this same path and being more aggressive driving traffic to the websites. This is a great example of how you can go about actually understanding what the value of a website is, and really measuring that, quantifying that value using real customer data that is tied to purchases.
I am going to back up for a second and talk about some more history. This is Vilfredo Pareto. He was an Italian economist. He lived at the turn of the last century, and you have all heard of the 80/20 rule, or the Pareto's law, named after this guy. And, what he was basically interested in was global wealth distribution. He came to the conclusion that, lo and behold, the vast majority of the world's wealth is concentrated in a very small percentage of the population. (Well, I could have told you that, especially since I lived in rural Bolivia for two years with indigenous farmers.) But, anyway, if you think about that in the context of what is going on in marketing today, you find that across all kinds of brands it is a very small percentage of households that are driving the vast majority of sales, and from which the major marketers derive their profitability.
So, in the case of Tide detergent, for example, it is a brand that has been around for 55 years, but, in reality, only 5 percent of households drive 60 percent of sales and 70 percent of profit. I have these numbers in my book. It's true across all kinds of brands.
I was sitting with Jim Stengel recently, the CMO of P&G, and he said, "Yeah, this is absolutely right. And, what we need to do is reallocate more of our marketing dollars toward the people from which we derive our value." And, you know A.G. Lafley has said the same thing publicly. So, what did they do last year? P&G reduced its spend on TV ads for Tide by 16 percent. And I really think that is just the beginning. They are beginning to realize that, even for a brand like Tide, which you think is a mass marketing product, it makes sense to understand, "Who are the consumers that account for most of our value?"
Tomorrow: Context, and marketing smarter.
