What Good Looks Like: Data Rich, Talent RichPosted: December 1, 2011 Filed under: Measurement and Analytics | Tags: analytical skills, data mining, innovation, Lean Six Sigma, monte carlo simulation, P&G, productivity, Robert McDonald 3 Comments
In an earlier post (The Good, the Bad, and the Ugly) I shared my perspective on how to think about degrees of readiness for Lean Six Sigma and, given the current level of maturity of an organization, possible modes of deployment. Sometimes it can get a little discouraging working in or with organizations that do not value data-driven analysis. But this example provides perhaps a bit of light at the end of the tunnel.
A recent interview in the McKinsey Quarterly highlights an organization that is rich in data and has a leader who understands the value of analytical skills. Michael Chui and Thomas Fleming of McKinsey interviewed the CEO of Procter & Gamble, Robert McDonald, and it is quite clear from their interview that McDonald is a great example of the kind of leader who understands the strategic value of information and advanced tools to wrest insight from data.
In their article McDonald says:
Our purpose at P&G is to touch and improve lives; everything we do is in that context. With digital technology, it’s now possible to have a one-on-one relationship with every consumer in the world. The more intimate the relationship, the more indispensable it becomes. We want to be the company that creates those indispensable relationships with our brands, and digital technology enables this.
One way is through consumer feedback. In 1984, when I was the Tide brand manager, I would get a cassette tape of consumer comments from the 1-800 line and listen to them in the car on the way home. Then, back at the office, I’d read and react to the letters we’d received. Today that’s obviously not sufficient—you’ve got blogs, tweets, all kinds of things.
And so we’ve developed something called “consumer pulse,” which uses Bayesian analysis to scan the universe of comments, categorize them by individual brand, and then put them on the screen of the relevant individual. I personally see the comments about the P&G brand. This allows for real-time reaction to what’s going on in the marketplace, because we know that if something happens in a blog and you don’t react immediately—or, worse, you don’t know about it—it could spin out of control by the time you get involved. The technology also lets us improve things that are working. For example, we’re rolling out a product called Downy Unstopables, a fragrance addition you can add to your wash, and the real-time comments from consumers about the product’s characteristics are helping us figure out how best to join in the discussion through our marketing efforts.
First of all, it is refreshing to hear a leader use the phrase “Bayesian analysis” (when was the last time you heard a CEO mention Bayes Theorem)? I can’t count how many times I’ve seen senior leaders tune out and consider various analytical approaches as too complicated or impractical. In most cases, these were managers from organizations whose cultures and management did not understand or appreciate what advanced tools could do for their businesses. But more damning, because to be fair most of us rarely have to use or understand Bayesian analysis, they lacked the curiosity to make an effort to learn and to support others in their organization to learn and apply analytical tools.
On the possibilities for operational improvement, McDonald adds:
From an operational standpoint, we also believe that to be successful we’ve got to continue to improve productivity, and being digitally enabled allows for that as well. So we’re digitizing our operations everywhere—from our manufacturing plants to the stores where consumers purchase our products. We believe digitization represents a source of competitive advantage.
In our manufacturing plants, for example, we have systems that allow people to use iPads to download data off the production line in real-time and communicate that to a place where we roll the data up. We’re not there yet, but we envision a system where I could literally see, on my laptop, any product at any moment as it goes through the manufacturing line of any one of our plants. And what I’d love to be able to do is see the costs of that product at the same time. It’s challenging because accounting systems aren’t designed today for operations—they tend to look backward—but we’re working on integrating our operational system with the financial system to move in that direction.
Another thing we do is to use our scale to bring state-of-the-art technology to retailers that otherwise can’t afford it. Imagine a small store in the Philippines, for example—a country where I used to live. We can provide sophisticated ordering applications to help people there run their businesses better than they would be able to otherwise. We have mobile-phone applications that allow retailers to order from us wirelessly or, if they don’t have a wireless capability, to order when they go back to their office and set the phone in a base. It’s very easy to use.
We also have performance standards that retailers in developing markets can visualize on their phones. For example, we believe you should arrange your store in a certain way to maximize consumer sales. If you have a store that partners with P&G on this, you can call up the performance standards on your phone, hold it up, look around your store, and compare it with what you see. Eventually, I want to be able to take a picture of the shelf, have it digitally compared, and then automatically send action steps back to the retailer to help rearrange the shelf for maximum consumer sales. That’s where we’re going.
McDonald’s description of how they are using advanced simulation and modeling is a good example of what good looks like in terms of an organization that values data and also has the managerial maturity to develop and recruit the talent needed to exploit that data.
Data modeling, simulation, and other digital tools are reshaping how we innovate. The way we used to do innovation research required a lot of work and time setting up consumer panels—you need the right distribution of races, ages, and so forth to make them representative. Now, with the amount of data we have available, the “n” is so large that by definition we can immediately have a representative group.
When you design a disposable diaper the traditional way, for example, by the time you get to the point where you make a prototype, the prototype itself has cost thousands of dollars, if not more, and it was all made by hand. Now, using modeling and simulation, you can go through thousands of iterations in seconds. The key is that you’ve got to have the data. So the advantage for P&G is our scale. We have operations in around 80 countries, our products are sold in almost every country, and we touch more than four billion consumers every day. Imagine all those data points. We can literally fit any virtual diaper to any baby anywhere in the world.
We’re even digitizing the creation of molecules. For example, in the research and development for our new dishwashing liquid, we used modeling to predict how moisture would excite various fragrance molecules so that throughout the dishwashing process you get the right fragrance notes at the right time. We did that all virtually.
At P&G they clearly understand the need for data that reflects “the-here-and-now” rather than what happened last week or last month, as is the case with many organizations who focus executive time reviewing cost variances against standard from last month. For example, McDonald describes below the inherent problems of store data collected that is two months old; this is an example of the kind of forward-thinking required to make best use of data, rather than looking in the rear-view mirror:
Similarly, every Monday morning we have a meeting with our leadership team all over the world—physically and virtually—where we review the business for the previous week and click down on all this data. And everyone signs up for the principles behind this—it’s real time and continuous; it gives us the ability to click down to find causality, make decisions, and then move on.
As we apply those principles each week, the challenge becomes the data source. I’ll use the Philippines again as an example. If a company we buy syndicated data from goes into stores in the Philippines once every two months and does a handheld questionnaire audit, then it doesn’t matter if we meet every Monday or not. Our data’s not going to be very good. So we’ve been working with all our data partners to help them understand that our need is for real-time data. For us it’s really constraint theory—understanding where the constraint in our data is and pushing it all the way to the data source. Then, change the data source.
For companies like ours that rely on external data partners, getting the data becomes part of the currency for the relationship. When we do joint business planning with retailers, for example, we have a scorecard, and the algorithm is all about value creation. Getting data becomes a big part of the value for us, and it’s a big part of how we work together. We have analytic capabilities that many retailers don’t have, so often we can use the data to help them decide how to merchandise or market their business in a positive way.
Perhaps the most refreshing comments from the P&G CEO revolved around their appreciation for the need to recruit, retain, and develop analytical talent. There is a great spot in the interview when McDonald asks a group of employees “How many of you have coded BCD?” or, “Have you ever done a Monte Carlo simulation?” To have a CEO who understands BCD (binary-coded decimal, a 4-bite representation of 0 to 9) or has an awareness of Monte Carlo modeling is an indicator that an organization has the top-level support to build and foster analytical skills as well as having a role model for setting the tone for all managers.
It is not very important whether P&G has a “Lean Six Sigma program” because it doesn’t matter what label one puts on things. The more important aspect is that P&G clearly understands the vital importance of how advances in digital technology can fundamentally confer competitive advantage in terms of productivity, execution, and innovation on those firms willing and able to leverage data through advanced analysis.
It would be heretical in this company to say that data are more valuable than a brand, but it’s the data sources that help create the brand and keep it dynamic. So those data sources are incredibly important. Therefore, we go to the extreme to protect whatever consumer data we get. It’s a board-level enterprise risk-management issue for us. We have very clear firewalls between one retailer and another and strict policies—for example, about how long a “cooling off” period you need to have when working on projects with different retailers. All of this comes with our strategy of being the most digitally enabled company in the world. We can’t do that without being an industry leader on data security and privacy.
When I started with P&G, in 1980, almost nothing was digital. Back then, our Management Systems Division—as we called it then—had mainframe computers, but our people did more work on phone systems than on computers. And whenever I would get together with them, I would ask, “How many of you have coded BCD?”4 or, “Have you ever done a Monte Carlo simulation?” Nobody would raise a hand. They didn’t have those kinds of skills.
More than two decades later, as vice chairman of global operations, I and my colleague Filippo Passerini, who today is the CIO of P&G, began to put together some very clear strategies to hire people with different skills. We needed people with backgrounds in computer modeling and simulation. We wanted to find people who had true mastery in computer science, from the basics of coding to advanced programing. When you’ve actually done a simulation, you truly realize the importance of the data; it’s classic “garbage in, garbage out.”
We’ve come a long way toward meeting our goals today, but we still have further to go. For example, we established a baseline digital-skills inventory that’s tailored to every level of advancement in the organization. We have a training facility to make sure that if you’re in a particular area, you’re competent on the systems for that area. This goes for senior managers too; we have an area in the facility where we can pull the curtains, so to speak, and work with senior managers privately so we don’t embarrass anyone. But we’ve got to have the standards for everyone because otherwise we’ll dumb the organization down to the lowest common denominator.
Ultimately, though, P&G has been pretty good about hiring for analytical thinking. We hire very good people and then train them. I remember the day I joined the company and one of the managers a few levels up said, “Throw away your MBA textbooks and we’ll teach you; we’ll give you another MBA.” And I think that’s still practical and relevant today. Nonetheless, analytical-thinking skills have become even more important to this company. We need to come up with the ideas to innovate, and those innovations are always informed by data.
Great Article Bruce. Got me thinking as usual.
“what good looks like: data rich, talent rich” it is very deep since in one page, I have found all ingredients and concepts , strategies that require for a company to outperform competitors. I will name some that strike my left and right brain:
” leader who undertsand the value of analytical skills”
” understand the constraint in our data”
” real time data and data source”
” value creation”
” integration of operational system with financial system”
” data modeling , simulation, and other digital tools are reshaping how we innovate”
” we have got to have the standards for everyone because otherwise we’ll dumb the organization down to the lowest common denominator”
‘ we hire very good people and then train them’
The list can go on and on. In the other hand, it will help to understand what are the drawbacks on relying on too much data-driven analysis. How much data is enough? I have experienced sometime we get lost in deeping and analysing every aspect of data which may not create value. sometime people procrastinate and they find excuses due to the lack of sufficient data.
The second question will be : how long it will take to bring a company to this this stage : “data rich, talent rich”? last night I was reading a paper by Roger E.Bohn, about Measuring and Managing technological knowledge( Sloan Management Review, MIT Fall 1994, volume 36, number 1). when applied the criteria by Roger E.Bohn , I may say “Data rich , talent rich” is the Nirvana stage. How achievable is this target?
Hi Denis – my comment on your second question is that it is indeed achievable; P&G’s financial success historically and its prospects going forward are, I would argue, evidence of the fact that some firms can reach this level. But as with many things, there are probably a few firms that are in this state and many firms that are not, just as there are many forms that outperform the market and many that are at or below the mean.
Part of the ability to glean insight from big data is in fact the ability to apply judgment and risk management to analysis. Put another way, firms that are talented at analyzing and mining data are by definition good at filtering out noise, focusing effort where the potential payoffs are huge or where the downside risks are also large. In fact in my view a company that has managers or workers that get stuck by analysis paralysis are by definition part of the lower right corner.
I’ve also seen some firms make some good progress toward the upper right corner but then start to regress. Often this occurs when a new leader takes over who is recruited from the outside and has a different set of priorities for the company, or when the organization becomes more short-term and cost focused and as a result cuts the kind of R&D and analytical talent needed to build future success.