Measure for Measure
Q&A with Dave Clementz, former CIO of ChevronTexaco Corp.
What wasn't right before about the way the company accounted for what IT contributed to the business?
We had to get everybody to bill their time, but we didn't have a way for them to do so. We had to come up with an electronic time sheet and it had to be easy to use and viable. Then we had to be able to roll up everything at the end of the month to determine, around a given product or service, how much labor was actually being used. We didn't have those tools in place. All we had were traditional, functional buckets. For example, I'm in the applications development group, and here's what my group's costs are. I have to figure out how to distribute that cost versus what the group is working on and what they're trying to do. And then I have to build an accounting system that allows me to track all costs associated with that particular product or service and try not to smear it all over the place, like a peanut butter sandwich.
How did having better measurements in place help to close the alignment
gap at ChevronTexaco?
There are two things we did to improve the alignment. First, we set up portfolio management. Now, once a year in the spring, we go out and talk to the customers about what they're going to need for the following year as they build their budgets for their expenses. And as we're building our budget for internal use, we need to establish what they are trying to do-how many phones they are going to have, how many desktop computers and so forth.
Now, my portfolio managers go out and talk to the IT managers. In corporate finance, someone looks over charges for the ERP system. So we now go in and have a conversation with that particular individual. We ask them how many applications they're going to need to adapt next year, or how many new things they want us to write. We talk to the people who actually have the spending power in the operating unit, and it eventually rolls up to the operating unit head, who says yeah, okay, I'm going to authorize $90 million to be spent next year on IT for my operating unit. It rolls up to the operating unit head, but the conversation about the details usually goes through the IT manager.
How does that help alignment?
Before, the process was, here's what it costs me to run this shop, so here's what your charges are going to be next year based upon some arbitrary formula-either your revenue or the number of people in your group, or something like that. That's a very common allocation method. As a matter of fact, our HR department uses that method based on head count. Here's what it costs to run an HR department: You have so many people, so here's your share of it. That's the way it used to be with IT, too-but we said, no, we're going to figure out what the customers really want. Only then can we align our product and service portfolio with the roll up of what all the customers want, so we can make sure if they don't want something, we don't carry it and, say, outsource it instead. For example, we used to have our own IT training department, but classes were only about a quarter full and it was a big expense, so we eliminated it. We now outsource all the training and offer it on demand. But we didn't know that very few people were actually using it until we went out and asked the customers how many people they were going to have to train. Communication-actually talking to people and measuring costs based on demand and real capacity-is a huge help to building credibility as well as getting an accurate picture of costs.
So by moving to this portfolio approach, it was actually forcing people
to have conversations rather than assumptions.
Yes. We produced a large book, basically, that reflected what each customer or business unit was using-how many phones, how many desktops, how much mainframe MIPS, server farms, etc. We would ask: "Do you want to continue at this level of support next year, and are there other things we can do for you?" By asking them, talking to them, we'd get to know them, what they needed and what they would like.
For example, we can now tell a customer, say, that we understand they're trying
to do business process reengineering. With that understanding, we can tell them
we think we have some people who are really pretty good at that and can help
folks think it through and maybe create an e-solution. We can tell them we can
write an application for them next year and it will cost this much, and then
put that on the docket for them next year.
We also hold Business and IT Forums every year. We hold two each year, and they're very valuable meetings. One is held early in the spring, when all of the IT managers from the whole [ChevronTexaco] corporation get together in one location for a couple of days and they bring a few of their business partners along to the meetings. Usually, all my product and service managers are there, and these meetings are attended by, maybe 100 to 200 people. During the merger, of course, everybody and their brother was trying to get involved.
And then, for instance, we'll usually talk about some of the things that we see coming down the pike in terms of technology, tell them what we'll be been doing for them, show them the shape of our portfolio for next year, and, ask, is there any interest in wireless technology? Is there any interest in Linux? And so forth. We'll also have a couple of experts on hand who will talk on topics like emerging technology, and then well have breakout groups around all the different offerings. There could be a breakout group around wireless and one around something else. And in the evenings, we'll usually have what we call the technology showcase, where some of the technical folks will set up poster sessions, and we'll have a cocktails reception and people can wander around in the poster sessions in the evenings.
The Business and IT Forum in the early spring tells people what we're thinking about, what the shape of the portfolio is going to look like. At those meetings, we'll send out the portfolio to the operating companies.
And then we'll have a summer meeting, and at those we really get into the IT offerings for the following year-a discussion of rates, the transparency of cost for everybody, and so forth, so they can go back to their operating units, put their budgets together and sit down with the head of their operating unit and be very comfortable and confident in explaining what their IT budget was going to be.
Because of this activity-based costing, they have a clear line of sight to the product and services they are going to order. Those two [kinds of] meetings are very critical, and were especially important during the merger, to getting people and their needs known and getting the products and services available to meet those needs.
Now, what has been really exciting for me is the network these forums create, the people you get to know. I always used to say when I gave my opening comments at these forums that one of the primary purposes of people getting together was to expand their network, and I challenged every person in the room, in the two days they were there, to meet ten people whom they didn't know when they got there-at least ten. I asked them to try to get to know these people well enough so they could call them and talk to them and they could be part of their networks as they went out and did their work. Then, when we got into these big IT deployments, like the desktop deployment or some kind of server consolidation or whatever, this network was already in place. And people were talking about the issues and sharing best practices. These forums are more than just a one-way communication between the IT group to the business units and vice versa. These forums also are a networking opportunity that give us a lot of dividends later on.
How did you get this as-told-to information into a database to determine
what your 60 or so different business units really wanted, or would like to
Maybe a large operating unit would have a half-dozen or a dozen profit centers, so each one of those profit centers would send a representative to these forums. We didn't have to deal with every office. The global lubricants division of Chevron, for example, was in almost every country around the world. But they had a single lubricant CIO or IT manager, and that person would have their contacts in the different countries and the larger units would have their own budgets, but not everybody would. So they'd manage either with direct billing or some allocation. We'd gather this information from them. The portfolio would go out as an offering, as an electronic document that basically outlined the products and services they'd be getting the following year based on their history, plus anything new that we thought they'd be interested in. Then, they'd respond by sending that electronic document back to us. All of this would populate our database so we could roll up our costs. It went into these various cost budgets for the various products and services, and then we could roll it all up and see if gazintas is equal to gazadus for the following year.
Nobody likes change. What kind of grousing was there at first, and was this
a result of credibility gap that had been with IT?
It was very difficult at first. I think the first attempt at this Business and IT Forum was around the common desktop. Back in 1998, we were deploying 30,000 desktops around the world, this was Chevron only. The senior management had all agreed to it, and the next level of management had mostly agreed to it, but all of the people in the field were saying "Those guys in IT are smoking some bad weed from Sonoma or something and got twisted off and can't get it back together again." I'm joking, but we figured it was time then to get everybody together and go through all of these IT costs and so forth with them.
And so, at that very first Business and IT Forum, there was a lot of grousing. We actually created a voting technique so that people could vote in their block, all the upstream people would get together, and the downstream would get together, and all the international would get together. We gave them these colored cards to vote on various topics, and that's how we'd vote and decide on how we were going to price things and so on and so forth-what was going to be in the desktop, what wasn't, who was going to opt out and who wasn't, and whatnot.
There was a voting box?
No, we used colored folders. We had red, green and yellow. We said okay, what do you think about this proposal, and we'd put the proposal out there, and we'd see what color cards came up. Red meant "Stop, I don't agree." And if there were red cards, then we'd stop and hammer it out, right there in the meeting. We would stop and drill in on the table that waved their red cards and say, what's your issue? Let's solve it. And the whole group, together, could then decide on the issue. Example: Network charges. How are you going to charge somebody who's in Nigeria for the network? The original plan was to divide the world up and charge everybody a fraction of what it would cost to run the global network. It turns out the people in Nigeria don't have nearly the traffic on the network as the people in the U.S., so the red cards went up. And most of the international guys had that objection.
So right then, we got the issue in front of us. We didn't solve it at the moment.
But by the next day, the network guys had figured out a formula that they could
use that would give relief to the international people but meant that the rates
for the domestic people would have to go up-because they were, in essence, being
subsidized by the people who weren't using it very much. They had to agree,
so then red cards would start going up on the domestic side. And so back and
forth and back and forth, until we finally got the global view, which was putting
the corporate hat on and getting everybody to think that this network was actually
a corporate resource, and the people who use it the most should be paying the
That's an example of how we'd solve these things. Those early meetings were very contentious, but I think because we were willing to listen and were willing to act, we'd actually take their advice, act in real time and make some decisions. All my managers were there so we didn't have to go off and ponder things for months on end-we could make decisions right there in the meeting or overnight, and move on. And so I think that those first couple of meetings were really awkward and frustrating, but then they became much more collaborative and open. Yeah, change is not easy and it's a pain-and especially if you have an IT organization that doesn't have a lot of credibility.
The operating companies when I first came on basically had duplicated a lot of the services that were being provided by the enterprise IT organization, because they didn't trust the service quality or the speed with which things got done, or because the cost was too high or whatever, so they would build their own. One of my challenges was to try to claw back some credibility so they wouldn't have to do that. And in fact, during those early years, a number of business units gave the enterprise IT group their IT people, and said, "Hey, why don't you do this for me as a service? I don't need to run an IT department out here in this organization."
How many people attend these forums, typically?
About 150. Enterprise IT people, business IT people and some of the business unit leaders, the technical leaders, are all there.
This got everybody talking for the first time?
When we started in 1997, that was the first one that I'd ever heard of.
So now these forums are permanent?
Yes, they've become institutionalized. This last year, we tried something new because we were in a money crunch and people were trying to watch their travel budgets, and discretionary funds were tight. We tried the spring meeting as a virtual meeting. We put the portfolio meetings out there in a Web site, in a portal, and we let people wander through it and send e-mail questions and comments. But it was only marginally successful. I think people really missed the human interaction. So we followed it two months later with another meeting, just the high-level IT managers came in person, along with just a few of our portfolio managers to minimize travel, so that was only about 80 people. I think we'll go back to the full in-person forums next year. You need face-to-face communication to be successful with this stuff. It can't all be virtual.
So your push to start measuring the impact of IT at your company, with these
forums and through other methods, was not only about measuring levels of dissent,
but also getting into the specific numbers and metrics-like how much, how many,
why and what about next year, right?
It's all about running the IT department as a business. You have customers, they have needs, you need to forecast their needs, you need to build budgets, then you need to recover all the costs. In our case, we had to recover to within 97 percent to 103 percent of our budget. We couldn't overcharge, and if we got too high, if we were recovering too much, I'd always give rebates. I'd figure out why we were recovering too much. Maybe we hadn't been able to reduce our rates, for example. Giving customers rebates gives the IT people a lot of credibility.
When we first did the activity-based costing, I discovered the mainframe was over-recovering by about $30 million a year, and $10 million of that was being charged to the marketing department. So I went to the marketing VP and I said look, I've just done an analysis of the costs, and I see that you've been overcharged for your mainframe, so we're going to reduce your mainframe charges by $10 million next year. He looked at me like he wanted to kiss me. He operates on a real thin margin. So I said, well, have fun building another couple of service stations. But he said no, what he'd actually want is to take the $10 million and invest it in a Web solution to interact with their private investors, the franchise holders. The project they started ended up costing about $10 million to provide this e-commerce solution to allow their jobbers and dealers to interact with mother Chevron. And he took the money, the savings, and reinvested it in IT. He didn't take it off and build more service stations with it. The credibility when you do that sort of thing is phenomenal.
How did you get that kind of precision with your metrics?
We had to build the accounting system to give us that. We had a portfolio of services that we were providing when I took over. We knew we were providing mainframe, server farms, desktop and network and so forth. Each one of those major product lines had a cost associated with it. In the old days, it was all kind of blended together. I had to hire a bunch more accountants, but you can hire more accounts with $80 million-what we saved in the first three years-and so I took each product line and broke it down with enough granularity, like how many people, how many software licenses, how much hardware was related to that one product or service and then took it down again for the individual customers. In order to get that degree of precision you need to have an accounting system that breaks it down by the activity.If I'm going to provide a network monitoring service, what does it cost me to do that service? I need so many servers, people, and I can roll up all the costs related to that one service and I say okay, now I have the granularity that I need. It's not rocket science. But it is tedious.
It took us six to nine months to get the accounting system to where it could handle it, and then another year to a year-and-one-half of change management to get the people to actually get accustomed to it, and then to get them to start billing their time.
And you had to handle more accountants.
Yes. But an accounting department is not expensive relative to the savings you get when you're able to do this. Accountants aren't that highly paid, so if you have your senior technology people who aren't that expensive doing this kind of work, that's the wrong answer. If you can build up your accounting department, it definitely pays out and if it doesn't, you can always get rid of them later.
So the IT credibility gap at ChevronTexaco is closed now?
It's a constant struggle. Service levels are moving targets. These days, as more and more people put more of their business on the digital backbone, there are more and more demands on IT. Last year, we created a metric around business incidents, the dollar value of (IT equipment) outages. We would measure every outage that happened, and for each, we'd try to work with the business units to develop the cost of that incident. So when the Melissa Virus hit, how many systems went down, how long did they go down and what was the potential revenue loss because of that or productivity loss because of that? We had a process we built that allowed us to get that kind of detail. Every one of these measurements ebbs and flows. Sometimes, you might be measuring one thing but that's not what really matters to the customer.
We had one guy whose system went down; it was a trading system. It was as serious as an oil spill because of the potential risk to the business and the damage to his operation. He was trying to make a point that you guys shouldn't be taking these things lightly. We weren't, of course, but system reliability is very important. In this case, the primary system failed and the backup failed within minutes. It was unheard of. Now we have our IT people conduct a business incident review and it invites business-side people to fill out a template to calculate the impact of the outages.
Generally speaking, why don't more companies measure costs and spending
more precisely? What observations can you share?
We don't want to crow too much about having all these metrics, but whenever I'd take my story on the road to a group of operating company people and show them how we grade ourselves and how we made reliability one of the metrics upon which the senior people in the organization would be judged on their bonuses for the following year, their eyeballs would pop out of their heads. They'd ask us, "You guys really ding yourselves?" We showed them the cost recovery demands, that we have to recover our cost within 3 percent or our bonuses are affected by it. We have safety metrics. We're dinged if we don't meet certain safety targets, and so forth. It gives you a lot of credibility, and with your own people, as well. Why don't a lot of people do it? Maybe it's not knowing what levers to pull. We definitely had a credibility issue when I first came on, and the only way out of it was to make the operation much more transparent, starting with the costs. Why does it cost this much for that? And what happened was, when you got enough detail to know how much it cost, you can say well, gee, why am I doing it that way? I can do it this way and save a whole bunch of money. Or, I've got two groups doing the same thing. Why do I have people doing the same thing? And that's where we started getting the savings in eliminating the redundancies, and we could do that by creating transparency. I think a lot of companies just grow so fast that they don't have time to think about what the important measurements are. The system itself becomes the measure. Did I get the system in on time or within the budget that was allowed? The easy things to measure are how fast did it get installed, and how much did it cost relative to what the budget was. And then if people are growing, they just focus on those things and they don't really stop to think about system reliability or security. We have something we call a penetration audit. We had some serious issues with the network and its vulnerability when I first got there. But if you don't measure it, how do you know where you are?
The Role of Standards in Cloud Security
Security is often cited as a primary cause for concern...Watch Now
Ensuring Resources for Mission Critical Workloads
Application workloads can thrive in cloud environments,...Watch Now
Improving Security in the Public Cloud
One of the main concerns about moving data to a public...Watch Now