Six Sigma Marketing

Jim answers questions from fellow Drillers
(More questions with answers here, Work Overview here, Index of concepts here)

Topic Overview

Hi again folks, Jim Novo here.

Six Sigma Marketing is really a “mindset” rather than any particular tool or technique and the example below touches everything from web metrics to the value of real-time analytics to customer retention to why engineers are trying their hand at marketing.

The core of my take on it is this: I don’t think marketers should be looking for defects in “marketing programs”, as some suggest.  I think Six Sigma Marketers should be looking for defects in the resulting “products” of these marketing program “plants” or “lines” – the customers they create. 

Anyway, I hope you enjoy this Q & A, and any comments on the idea of Six Sigma Marketing are welcome, particularly from engineers.

Let’s do some Drillin’!

Q: In the few years I’ve spent working in the field of digital analytics I’ve only met a very small handful of companies that are able to take immediate action based on data. Because of this, I’m generally less excited about “real time” data than “getting the right data in a reasonable amount of time”.

But more people seem to want detailed / enhanced real-time digital analytics, which are of course more expensive.  Is there a case to be made for this idea, and how would you go about justifying the additional expense?

A:  There is a profit opportunity in virtually every business for real-time analytics, but the opportunity will probably be very different for each business.  I think this idea is so important that I’m going to spend a significant amount of time not answering this question specifically, but using a “business model” to explain it.  I really want people to think about the following and then relate it to their own businesses.

Engineers know about real-time, they live it. Ever seen thepictures of a NOC (Network Operations Center) for the telcos or the electric utilities? These engineers run the guts of the business in real time, they are sitting there in front of a field of screens that alert them to trouble and they react more or less instantly, or they monitor / correct the action taken by a robot. At a lesser scale, in every company there are IT people that walk around with alarms that go off at all times of the day to alert them to processes and equipment that have gone haywire. My wife was one of these people. It’s real-time monitoring and alerting, it’s here and it’s being done all the time – in engineering and in IT.

It’s obviously important to the business, or they wouldn’t do it, because it is pretty damn expensive.  So why do they do it? Because there is a clear idea of the value of it.  If business processes go down, the company seizes up and there are clear ramifications to this, there are sales not booked or costs incurred that happen “in real-time”.  So the monitoring is in real-time, and it works.

There are amazing success stories in some other internal / operational areas – transportation / logistics, for example, where again, the benefits are pretty clear, because the impact is immediate.  Much of manufacturing is now done with real-time monitoring.  But these are the “hard sciences”, they are internal, engineering and numbers driven and engineers understand real-time.

So I think the question behind this question is, what does this real-time idea look like when it is “exposed”, enters the realm of the “soft sciences” like marketing, has to be “external” and customer facing, has to be used by people who are not engineers?  People who have never even thought about doing their job in real-time, never mind executing in real-time?

See the average web site.

Let’s take a more macro view of this, talk about the “junction of engineering and marketing”.  Engineering is a real-time hard science.  Marketing is not, at least most of what people would call marketing.  Web analytics and the web in general suffer from the problem of sitting at this engineering / marketing junction, web sites are about engineering and marketing, both at the same time.  The challenge is, marketers don’t understand what the engineers are saying and vice versa.  The engineers ask, “What do you want for reports?” The marketers say, “What have you got?” The engineers say, “We can give you anything you want.”  The marketers say, “Give me what is important.”  And this is how you get stacks of web analytics reports that nobody uses for anything.

The funny thing is, there are legions of engineers and IT people who have decided to “do something” about this problem by “learning marketing”, but there are few marketing people willing to “learn servers”. What kind of marketing are engineers and IT people willing to learn?  Think about it.

The only kind of marketing that makes sense to them – direct / database marketing.  The only kind of marketing that can be measured, automated, turned into an engineering project – complete with ROI.  The implications here are staggering – and this is also why there is so much “bad marketing” around on the web in general.  Because engineers are not good marketers, but they have no choice, because most marketers know next to nothing about database marketing.

For example, Pay-Per-Click bid management systems. They spider the listings and adjust bids in real-time, optimizing for ROI. Do you think a marketer came up with these? Come on. Most marketers initially thought search marketing was all bull. Engineers who knew something about marketing, about direct marketing, came up with these systems. They invented managing bids by daypart, which is a totally fantastic idea, because (as direct marketers know) the audience changes throughout the day. And whose idea was this? A marketer? No. Engineers, as truly “scientific method” types, discovered through testing that you could get higher ROI managing bids by daypart. They were basically “correcting defects” in the system from an engineering perspective. This process of reducing defects has been formalized and at the highest order is generally known as “Six Sigma”.

The challenge of real-time for marketing is to prove the value of it, as the engineers do every day when their alarms go off.

In the case of marketing, most of the value comes from opportunity costs that on the surface can be difficult to measure.  The defection of a customer, for example.  When did it happen?  How do we know when a customer is no longer a customer?  How can we slow or prevent a customer defection?

Does your company measure defection rates?  Or are all customers customers “forever”, unless they call you up and say “I think you suck, I’m no longer a customer.”  You have all heard the stats on the value of increasing customer retention – costs 5x more to acquire a customer that keep one, a 5% increase in best customer retention drops a 25% increase to the bottom line, etc.  But in the vast majority of  companies, nobody in marketing does anything about retention.  They don’t even measure customer defection.

That’s because customer defection is a database marketing concept, and most marketers know next to nothing about database marketing.  The fact is customer defection is largely caused by “defects” in the way the customer was “manufactured” in the first place, or “defects” in the way they were “maintained”.  And I know more engineers that understand these concepts than I do marketers who understand them.

And it’s because, literally, marketers don’t think like engineers.  If you explain opportunity costs, customer defection metrics, and the potential for increased profits to an engineer, they understand it immediately.  Because it’s about numbers, ROI.  Measurable.  It’s working or it isn’t.  It’s database marketing, it has a start and a finish.  Alpha and Omega.

In other words, database marketing is “engineer’s marketing”, it’s Six Sigma Marketing, and you can approach most database marketing challenges the same way engineers approach their Six Sigma challenges.

The real-time challenge for “soft science” or “customer-facing” business units, and marketing in particular, is to understand the concept of “opportunity cost”, the costs that are not so obvious, lost profits that occur in the future because you didn’t take action today.  Engineers understand these costs, because they know that defects today mean lost revenue / increased cost tomorrow.

In other words, these soft science business units have to learn how to manufacture customers correctly, with a low error rate.  And they have to learn how to spot customer maintenance defects remotely, anticipate defects in customer value.

Opportunity cost is like the opposite of conversion rate, it’s measuring what did not happen.  What percent of visitors who bought more than 90 days ago did not buy in the past 30 days?  Why?  These customers must have been manufactured incorrectly, they must be defective, because there is this other group of customers, manufactured by a different plant (campaign), who bought more  than 90 days ago and are still buying now, as Recently as yesterday.  So, what are you going to do about these defective customers?

The reality is it’s probably too late to do anything profitable about them, because you weren’t measuring these defects, so you couldn’t manage them.  You don’t have “trip wires” or “alerts” set up to “page you” when a “batch” of defective customers is discovered, you’re not even looking at any customer defection metrics.  So then two things happen:

1.  You probably continued to manufacture defective customers for some time longer than you should have, beyond when this “customer manufacturing line” (campaign or product) should have been shut down.  This is a waste of resources, time and money.  Opportunity cost.  Money lost by failing to allocate marketing resources to the best use, to a campaign or product that manufactures higher quality customers.

2.  It’s too late to take action to “repair” the defective customers with a follow-up campaign or action that attempts to correct their flaw.  These defective customers are now in the distribution chain, where they will remain defects and contribute no profit or even negative profit to the company in the future.  Opportunity cost.  Money lost by not doing something about the defects in a timely way.  In customer marketing, what you do today has consequences for the future value of the customer.  Every contact can either increase or decrease future customer value.

Now, if you have lots of customers, products, and campaigns, you can’t possibly keep track of all of this, all the potential defects that might arise.  So you need a real-time system that alerts you to take action, the “trip wire”.  It uses defection metrics to scan for customers acting in defective ways, so the line that is generating these defective customers can be fixed, and the capital reallocated to another better use – a campaign or process modification that over time reliably generates Six Sigma quality, low defect customers.

But here’s the difference between Six Sigma for Engineering and Six Sigma for Marketing, and why database marketing concepts can be generally difficult for people to understand.  In manufacturing, IT, or any of the hard sciences, real-time defect alerts tell you something is happening now, and you need to act now.  A robot sees a defect and stops the line immediately.  Then the engineers try to find out why this defect happened and generally reduce these defects over time.

Unlike the engineers, in marketing and the soft, customer-facing sciences (customer service, contact centers agent desks, sales personnel, etc.) the transactions with customers occurring now create defective customer value in the future.  When a customer has a bad experience now, this result does not affect customer value now, the “value defect” occurs in the future, as the customer reduces business with the company or defects.

So unlike the engineers, you need to act now to reduce defects in the future. This is because the value of a customer, external or internal, is a function of customer behavior over time, and what you do now affects the value of the customer in the future.

When business is all basically conducted face to face, this is not an issue, you deal with it “in real-time”.  But when you’re talking about doing business remotely, you’re talking about web sites and self-service and the like, you don’t really know what is happening, and the customer defects play out over time. And you don’t know about them because in many, many cases, nothing happens – and that is the point.  You have to measure what did not happen to see these customer value defects, and the majority of people, including engineers, are not used to this idea.  They measure what happens, not what doesn’t happen.  “Doesn’t happen” is generally good in the hard sciences (as in “width not outside acceptable range”), but “doesn’t happen” is generally bad on the soft side, because business is about things
happening – sales, visits, downloads, etc.

So it is one thing to monitor the initial conversion of 30 simultaneous campaigns in real-time, and make adjustments on the fly.  Some people do this, certainly in pay-per-click, either in person or using robots.  But this activity is happening now, it has value now, and you are doing something about it now; it’s manufacturing, it’s engineering, that’s “easier”, it is a process that is very well understood and easy to prove the value of because defect discovery and correction occurs “in synch”.

But what if I was to tell you that the pay-per-click campaign set you are optimizing right now is creating customers with very low value in the future. Further, the keywords with the highest conversion rate generated customers with the very lowest or even negative value in the future. In other words, these keywords were creating defective customers, relative to other keywords. Would that knowledge affect what you were doing now with optimizing the campaigns?

Sure it would.

And how did I know these keywords were creating customers with low value in the future? Because I just got pinged by the real-time analytics application, telling me that the customers generated by the keywords in question for the past 60 days had an abnormally low “likely to buy” score relative to the rest of the customer base. These customers are quite unlikely to buy again, they should not be manufactured any more.

They’re defective.

And what would you do?  You’d immediately (as in right now) lower or kill those bids and use the money to raise bids elsewhere, right now.  Because not to do it now would be a waste of time and money.  The marketing capital should be reallocated to a higher and better use – keywords that generate customers with higher “likely to buy” scores.

The longer you wait, the more money that is lost.  If you found this out 60 days later, when you ran “current time” analytics and found out they in fact had not bought again, you have wasted a ton of money.

It’s obvious, right?  And why is it obvious?

Because the real-time analytics and metrics have turned the future into now for you, and you can act now, even though the action has value in the future.  And as you allocated the budget money away from these keywords and towards keywords that create customers without these defects, you would be basically looking at your marketing through a set of Six Sigma glasses.  You are reducing defects that you know are going to hurt you in the future.  That’s Six Sigma Marketing.  And it is also real value creation by a marketing department using real-time analytics.  It’s the kind of value creation that can be precisely measured and used to justify the expense of real-time analytics.  Just like the engineers do.

Now think of programs like this throughout the company, each handling some kind of important customer-facing issue, each being optimized in the same way – to maximize customer value while at the same time driving the highest and best use of capital.

Now, do you have to use real-time analytics to be a Six Sigma Marketer? Not at all; you don’t have to use an automated system to alert you and your business doesn’t even have to be online. After all, this general approach to managing customer value was developed offline decades ago using the telephone and direct mail.

How “fast” defect detection and correction need to happen depends on two things: the natural rhythm and speed of the business area, and the financial stakes involved with the defect.  The faster the pace of the business and the higher the financial stakes surrounding the defect, the more money you will lose by not correcting defects in a timely way.  It’s probably a good idea to simply start with finding the defects in your customers.  You can worry about “speed” and all the rest of it when you have something tangible to work with.

Take the first step to becoming a Six Sigma Marketer.  Measure your customer defects, and get committed to reducing defective customer-facing transactions.  Your objective is to simultaneously increase customer value and reallocate capital to highest and best use.

That’s a concept every CFO understands.

Get the book at

Find Out Specifically What is in the Book

Learn Customer Marketing Concepts and Metrics (site article list)

Download the first 9 chapters of the Drilling Down book: PDF