One view of data, one decision and one way execution focus is the goal of every enterprise. A global single instance presents an IT option of creating the one view.

Abstract

In the business world where geographical boundaries are meaningless, companies always try to have a single way of doing business across all their business units around the globe. The very into the game panelists here share their views and experiences on global single instance. The session explores every aspect of a single instance - from business drivers to costs impact to disruptive forces in the marketplace to the technological aspects. A clear understanding of the business processes in different regions and countries and strong governance mechanism are among the key success factors that are being recommended for having a single instance.

Discussion

Greg James (On global single instance that Fonterra is trying to put in place): The dairy industry in New Zealand is a corporate dairy industry and was deregulated in 2000. Effectively this meant we were forced into a situation where we had a merger of three major companies and as we went through the initial journey of that deregulation and the entity that became Fonterra, we quickly realized that all those three major companies did things differently - different processes, different systems and different cultures.

As part of our journey, we realized that the model was not going to be sustainable and the only thing that it would sustain was probably a very rapid exit out of the global market. To retain our lowest cost manufacturing status, we needed to look at our business model. Hence we embarked on a program which started first as being business-led. The program was sponsored by our CEO and we started by a looking at a high-level business model. We then decomposed down into the various operating models that we wanted to put in place that underpinned that particular business model.

We dissected our supply chain end-to-end to determine how common our business was across various functions and geographies. We realized that nine times out of ten, the processes that we could deploy were actually common processes, they need not be different and the points of difference had basically been invented themselves over historic or over time, by people effectively doing things in different ways. So the key principle that we developed was that we would globalize.

From a business perspective, we had very clear views on what that meant. We would standardize on global or common processes across our entire operations and that principle was held steadfast. We recognized that by merging the various companies that we had, we had effectively replicated operations just about everywhere in the world. We looked from a process point of view and balance the way we wished to operate, we actually ended up with very fat structures effectively doing the same thing. So the key principle was our desire to move to a shared services-type environment for our back office & finance and administration functions and also for our customer service functions servicing multiple markets. In the end, we achieved that.

In the process, we learnt that we needed to have a strong master data management strategy and obviously we needed a technical strategy that underpinned the deployment of the global instance. From our perspective, that transformation was absolutely business driven. It was not an IT transformation. The businesses-led initiative sponsored by the CIO actually put 50 to 60 % of working effort over three years in this program. This was the catalyst for success, was the key, because the program is about business transformation and fundamentally it's about change management. We were prepared for those elements in the new era and the new way of working.

Jasmeet Singh: AIG Investments has multiple applications on ERP as well as a global single instance of implementation across several areas. Please expand on this.

Steve Weinreb: The three major themes or drivers are scalability, globalization and governance. What I mean by scalability is that we wanted to grow our revenue at 30 per cent plus per year keeping the expenses at a minimum or certainly not growing at the same pace which made the challenge even bigger. When it comes to globalization, I think all companies need to follow the marketplace. It becomes easier to manage if you have a common set of processes and unified data that you use around the world to ensure a level of uniformity and transparency. This seems to be a common theme in the financial services industry, not only within ERP.

CRM is important as all sales and marketing people see the same information on a client from around the world. They will make similar business plans around those clients. So that is important document management. Research or collaboration on creating documents for presentations is another case where building a common system actually yields a lot of benefits. It is important that senior management gets a view of data across the enterprise to make decisions. Every financial services firm now has a risk department and their goal is to manage that risk across enterprise. It is important to have information at your finger tips.

Jasmeet Singh: How do you implement something like this when you are undergoing a major change? AMD has undergone a merger or an acquisition, as well as a divestiture in that kind of scenario when there was a change that was happening. How did AMD go about that and can you comment on some of the things that you saw at AMD?

David Cefai: There is no question when you go through that level of change because it is going to add a certain amount of complexity through the projects. AMD is the other choice in the microprocessor industry. With the acquisition of ATI graphics about a year ago, we have a dominant position not only in the games sector but also in the high-definition television sector.

AMD's business strategy is built on three fundamentals. One is business innovation as demonstrated by our recent launch of a new chip which was the first true quad core processor as opposed to two duals wrapped together. Power efficiency is the key, we call it power per performance. Going forward, the strategy is built on generating the synergies between the microprocessor and graphic chips that's together and ultimately offering one robust platform and chips for our customers.

We are merging two different cultures here. AMD was very process-oriented, structured and planning-oriented. The ATI side was very entrepreneurial which is a politically correct way of saying they had no process. It was just spread sheet-driven and we realized that we would have successfully merged these companies and developed a strategy that needed to transform the way their business model operated.

I bring a little bit of unique perspective to this from my experience. I worked for an electronic manufacturing company that tried to achieve a single instance, never got there and ended up abandoning that part. I worked for a company that actually had a single instance strategy but it made no sense to them and they decided not to go down that path. In this case, it did make sense and the key point was a business decision that enables the company to achieve that strategy. We are about three months away from achieving that vision.

We've gone down our single instance path in three phases. We just completed phase two. The first phase was getting the financials onto a single instance. The second phase was getting the work management and inventory control portions onto a single instance. The final phase will address some of the procurements and remaining pieces of supply chain as well as some of our costing detail information. By Q1 of next year, we will be on a global SAP instance.

You have to look at business model and the strategy your company is trying to employ. Ask yourself do you have a single primary core business with similar business processes or a varied diverse process across the business units? Do you have strong central corporate governance or is it very decentralized? Does your business aspire to act as a single global company or a group of conglomerate of multinationals? You have to answer these to determine whether a single instance is appropriate or not.

Jasmeet Singh: There is a lot of talk on the customer having the control, about mass customization. With widespread consumer base, everybody is chasing and also trying to sell uniquely. Is there an inherent paradox here given that on the one hand we are talking about standardizing all the processes across a global single instance and on the other we are trying to sell uniquely and service uniquely our client base?

David Cefai: I don't think the two are mutually exclusive. In terms of the phase in which you provide it to your customer, you need to take a configuration-type approach and the key is configuration vs. customization. If you have done proper configuration, the upgrade should go fairly seamlessly. If it is tremendously painful, you probably have done customization. From that perspective, I don't see them as mutually exclusive. I think behind the scenes, especially for manufacturing companies, a lot of the core processes in terms of the supply chain management, the design aspects, and getting the product to market and getting the product to high volume manufacture ability -- these are all standardizable processes and those are the pieces that you try to take advantage of a single instance. We can be unique in terms of how we deal with end customers and whether we're dealing with an OEM or whether we are dealing with a distributor. We try to configure those touch points so they work, but they are all leveraging of the same back end.

Steve Weinreb: In financial services, it's all about data. It is really important, something like a common sort of accounts in an ERP-type implementation and getting the best practice in some ways. We went a long way to make sure that we didn't customize, we just configured it to meet the needs. So that is the right way to do it. In our outward facing applications, something like the Internet, intranet applications, we can make it a little more unique to suit the needs of each region.

Greg James: First, you need to understand what your business model is. Is it B2B or B2C? Different models have different demands. The change generally applies not in terms of getting the product to a point; it's where you're going to that final and ultimate customer. So it's about looking at your model, looking ahead in terms of how we can unlock the synergies and the benefits of the global instance in that B2B-type marketplace. Then the real challenge is collaboration. How we ultimately collaborate to take and add value to our top line as opposed to chopping away at the cost since that's just sort of what we should be doing as a part of the prudence in our management activity matters.

Jasmeet Singh: In terms of global consolidation of processes, you've seen HCL help clients move to a global single instance. What are you seeing in the markets? There are three industry representatives here, what are you seeing across industries?

Murali Raghavan: There are two types of things we are seeing. The ones which we are starting to implement or implementations for the last one or two years, have essentially started with a single global instance.

People who have implemented earlier especially in the packages scenario, diversified or multiple instances, have had that learning and most of them are trying to go towards a single global instance. However, the only caution is that a lot of backend processes are driving this global single instance or instances. Core processes or processes that differentiate or touch the customer are still a bit suspect as to how much of standardization are they able to bring, especially where there are diversified products, geographies which are different, fast changing company and governance.

We are seeing companies with such diversified offering looking at changing the core process which differentiates them. The most important factor is driving both the old ones, which are trying to become single instance, and the new ones. This is the single biggest factor which is driving these people to go towards single or lesser instances.

All this is either for the sake of decision-making or legal requirements such as Sarbanes Oxley. The second perspective is of instances which have already been implemented, multiple of them and the consolidation happening there. In a lot of cases it is driving towards lesser instances of rather than a single one. So it is an extremely difficult method of consolidation.

Jasmeet Singh: What are some of the business drivers, business results that you have seen which drove this? None of you have actually pursued this as an IT initiative.

Greg James: We did not have many options because we knew that the model we inherited when Fonterra was formed was sub optimal. When we went through the evaluation process, we dissected each part of our supply chain and identified, from a benefits perspective, we were looking at the long-term gains that we would get out of it. The benefits realizations when driven from a business perspective, actually becomes quite compelling. It is not about the savings you make by managing a global instance. This because while they are real, the benefits across our supply chain, inventory area - in terms of our inventory turn by having access to the information to make proactive decisions like whether to ship into Latin America or Southeast Asia as opposed to shipping to one place and having to send the vessel all the way to the other place which adds incremental costs. The benefits realization case was quite compelling.

Since we are a cooperative industry, we have a lot of regulatory requirements. We had our benefits ordered for by an external party. And we exceeded our benefits realization quite significantly. So our payback was a lot quicker than anticipated. The focus on business-driven, business-led business benefit is the key for a successful outcome.

David Cefai: At AMD it was quite interesting, the IT group was pushing for a single instance model for quite sometime. They had a hard time elaborating what the business value or business case would be. Our business partners just solely looked at this as an IT cost savings initiative and not a lot of people wanted to change. The driver for us was that we had to write off a half a billion dollars in inventory in one quarter because of multiple diverse processes and because of that we missed our quarterly earnings and got hammered by the stream. The board of the executives said we need to standardize the process, perhaps we should go to a single global instance. Our IT team agreed. In reality we got tremendously hurt writing off a significant amount of inventory because of these processes. We are only six weeks into our phase to go live, where we now have all our supply chain inventory processes on a single global instance. There are some growing pains as we go through that process but we are already seeing inventory accuracy rates at 99.8% and above. This has already paid back the cost of the project tenfold.

Jasmeet Singh: So it is not about cost savings or the cost savings that you get by consolidation it's more about business results?

Steve Weinreb: Those are nice benefits as well, but they are overshadowed by orders of magnitude in terms of business.

It's hard to quantify the values unless you have plenty of them. But I don't want to belittle it. The cost savings are all adding up because of Sarbanes Oxley and all of the logical user access change management, disaster recovery needs. But on the other hand you are paying more for the license for the software, so if you really start and the adding it up it is becoming much bigger expense item. I knew that the savings aspect it really was very big but that it is business driven again thankfully, because that is one benefit of Sarbanes Oxley and things like Avian Flu when it comes to trading systems. It's interesting to figure out a scheme that would save the whole region of the world or be able to process trading in another region of the world. What if you're on different systems then it becomes pretty difficult to provide that level of service. It all really starts to add up to a very solid business case across the board.

Question: Did you have a lot of trouble with cross reference tables to publish data across the applications?

Steve Weinreb: Yes. I'm not saying that any of this is easy. But because of multiple ways of doing it, you could actually build a whole SOA middleware around that. Or you could go to one centralized and pick out the common data elements that need to be shared, in our case its security master reference data. It is really important that anywhere in the world if you're trading the same instrument that you call it the same thing and label it the same way, otherwise it's going to be treated differently and when you merge into the books and records you have a problem. But then you get into some country-specific needs around statutory reporting etc. that's where you can get more into a model of a hand with the finger. So anything locally in the finger they can process they are outside the system, and anything that is supported has to come back into the hand can come back in, else it stays locally in that region.

Question: Once you've created this global instance how difficult or easy has it been to keep the data, operational data quality good?

David Cefai: It's much easier than it was before. It's always a challenge going forward when you've essentially eliminated one of the dimensions of complexity by going onto single instance, as well as when you're trying to achieve the goal of a single instance. A robust strategy around master data management is one of the prerequisites to get there. In many cases you have to address that problem before you can achieve the global single instance. Then you start reaping in the benefits. We have spent a lot of time in terms of resources that would get system generated reports. People would not trust data on the reports. Then they started developing ad-hoc data marks in the background and there was an endless reconciliation activity that was going on. If you are looking at resources who can do all of this reconciliation, you are looking at the delays in taking decisions and you're making critical billion dollar decisions based on bad data. So by forcing us to go to the single instance, we are forced to clean up a lot of that and put a strict governance structure to maintain it.

Greg James: If we had the chance to do it again even though we did it four years ago, we would focus a lot more on master data management strategies and on BI strategies. We're going to have to go back and start just to reduce some of those areas to get the optimization that we really need, but other than the business side of it, to me it is the two key areas that you can really focus on.

Steven Weinreb: The success rate goes beyond the data. But in the journey to go to a centralized model, the importance of change management cannot be any less highlighted. To implement our ERP system globally, we built a separate team dedicated to change management to make sure every region and every user was actually part of building that central model. You have to build your network of experts in every region, you have to have evangelists in every region, they have to be well-trained, need the documentation, newsletters, need to get around and build leghos together and other things that would make them all part of the central team. That's how you get this done in the long run.

Greg James: One of the key lessons that we learnt is not just to internalize. It is not just about the change management within our own organization that we had to put a huge degree of effort into both our supplier base and our customer base because we were going to a very different mode of operation. All the dimensions need to be effective.

Jasmeet Singh: Several disruptive forces are emerging in the marketplace and customer is becoming the dictator of products and services. Regarding middleware and SOA, a static ERP is a problem and this is getting replaced by dynamic distributed data and that is what is going to drive the success of any business which is based on consumers. So in a B2C model, is global single instance or global instance a drag on the business? Is it compromising the success of the business? Is it affecting? Is it impacting the agility of the business to succeed in a global market place?

David Cefai: By getting into a single global instance, we've become more flexible. One of the things that people often overlook when they are building their business case or during transformation is they are always looking backwards in terms of savings. They tend to overlook that business is changing and if you're on multiple diverse systems, you need to change the model, the process. The process change needs to happen across six different systems. So six times the time, six times the money to make that change. But now, since you are on a single global instance, you only need to do that once.

Greg James: We talked about 60% of data out there still sitting on mainframes. The difference I have is a packaged solution, my core processes are through that package, some enabled through the processes in that particular solution. If I take it a step further and look at SOA in my particular landscape, I have limited use for it, because I'm enabled through common processes. This is because customers are dictating the ways of the world and will continue to do so. A lot of these disruptive forces have been by the incremental and significant change in technology improvements over the last five to seven years. So we are creating those disruptions ourselves. We need to then come back and look at our business model, to minimize those disruptions.

Jasmeet Singh: In light of the middleware capability that exists today, for someone who's looking to move to a global single instance, does it make sense to move to a packaged application or patch everything together to utilize the legacy and still get the results?

Greg James: There are many answers to this question. But you need to dissect your business in terms of where your business stands, strategy and where your business would be in four or six years. You need to work out the commonality. If you have a business model that is absolutely different between your operating entities, then you would need to go to a specific business function instance time activity as opposed to a global instance time activity and then use your business intelligence. However, you might want to use it if there is commonality at least in some level of the data hierarchy.

Murali Raghavan: It's a matter of the way you architect SOA and the global instance is really driving an architecture which is capable of delivering services the way it is required. The packages offered are all driving towards this. There is no divergence; all of them are in middleware. Take for instance SAP's Netweaver and Oracle's Fusion. Whether it is any other middleware or their own, the drive is towards creating services under which it can be used. So I think there is no divergence in the view shared. The only point is how much innovation it gives. I think it does give with SOA, it gives a lot of flexibility. All of this is converging whether it is SOA or middleware.

Question: In moving to a single global instance, did you find that regression testing became a more complex and costly thing to do in enhancement cycles?

Greg James: We're used to HCL, so it's quite easy. It's a challenge which needs to be scripted carefully right up front so that you are getting into a repeatable-type situation. But the whole management around change is something that certainly needs focus.

Steven Weinreb: Regression testing, reconciliations, parallel runs across different systems all of this is a lot of work. I don't have any answer on how to make this easier, except that you need to keep a good plan and stick to it diligently. One of the complexities is that more and more packages are trying to do end-to-end solutions which gets in the way of where we're going. You have to use parts of those systems and then take the data and integrate those into your master data management systems.

David Cefai: One of the ways you can lessen the burden is by having a robust automated testing capability. The second aspect is that a lot of testing falls on your business partners as opposed to IT testing. We'll test the interfaces and make sure the bits that are going in are the same bits coming out. But the real heavy lifting needs to be done by your business partners. They are the supply chain experts, procurement experts, financial experts -- they're the ones who really need to ensure the system and the business logic is acting the way it's supposed to. If this is an IT-led project, then getting those resources to do that testing appropriately becomes even more difficult.

Question: Typically, when you bring out a new release, do I look for resources from every region and every territory to be involved in regression testing? That's my fear moving to this model since those resources are very difficult to find.

David Cefai: We'd rather experience that in the test environment rather than the production environment.

Greg James: In each area, there is a nominated person who will take responsibility, across geographies. When we start getting down into some of the specifics of some of the regulatory and legislative activity then we will tend to go into that particular geography itself and to make sure that we've got absolute coverage in terms of getting it right. But generally we have the same people that will get involved at those points on a regular basis.

Steven Weinreb: If you really need to get those people involved early on and that one of the things they are responsible for. Learning the systems and then ultimately being responsible for the final testing and sign off. So from a Sarbanes Oxley standpoint I'd have to produce that signoff anyway that they have tested their components so I would say it's really important that they are involved and they signoff otherwise I would not even do the project that they really need to be a part of it.

Question: The company I work for covers about 22 countries and we have a very federated model. There is a lot of independence and autonomy in each country of operation. This would probably be major governance mechanism, central governance mechanism for the group and quite a risky thing to do. I'll have to convince a lot of business executives to go along with it at this point of time. The next level that kind of makes sense of from a good governance point of view is the regional level. At that sort of level does it make sense to try to architect instances or deployments of application?

David Cefai: This goes back to not knowing your company and also the business processes in these various countries and regions. You need to quantify as best you can why you want to do this as opposed to just it is the corporate edict and for the real good you have to do it. But this argument doesn't sell. It has to be a process-driven discussion. If you know 80% or greater of the processes are similar, tremendous benefits can be reaped and those benefits need to be quantified to get to that point. But it needs to be a top-down driven thing. If you're trying to build consensus from the bottom-up, it is quite a long journey.

Greg James: Prior to our move in our distributed operating company model, even if you tried to do it regionally, you still need to get the consensus. Each managing director and each region was seeing it slightly differently. So the comment was: "My old system works well, why would I change it?" It's got to be a top-down focus and that's forcing the change and that doesn't necessarily mean there is complete control in command of activity. It still means you can operate collaboratively.

Steven Weinreb: I would ask is there any need at the consolidation level or the central management level if you have that type of structure in your company. That was a big driver in our case - it was central management, risk management, compliance and audit needing to see all of that information and data in a uniform way across. So if you are already getting those results then you may not have a business case. If you're not, then this is a pretty palatable approach. Less is better. Fewer systems to manage would certainly provide fewer moving parts and breakage points.

Question: We talked about when you've go to a single instance, there are master data challenges and that is a major change the company's IT plumbing. Have any of the three of you done the analysis on whether to do a major upgrade or any other IT activity, Should we package it or should you take this on as an independent project itself

Steve Weinreb: We actually did it. Separate pieces with each piece providing its own business case. It is hard to package it without it looking like a complete IT driven transformation. CRM has its own specific needs for sales and marketing and their ability to cross sell. Document management was driven by the research group so that they can share that data because we have research analysts in 44 countries. The ERP front had its own business case as far as getting uniformity and clarity of data across regions into our central reporting scheme. So those sold by itself and then the master data management piece in our case is almost separate and we're really in the middle of that now so we did the other three people individually.

David Cefai: We did it in pieces as well, more due to budgetary constraints. When you are talking of a big bang-type approach versus a sort of piecemeal approach, the piecemeal approach is more palatable for a couple of reasons. One you are talking about less overall risks to the company as well and I think you can start delivering business value quicker if you do it in pieces. There are some portions that it just makes sense to do it together for efficiency reasons.