LISA: You are one of the people with the most experience in this still relatively young business. How did you come to get involved?

Michael O'Callaghan: I was appointed general manager for Microsoft's localization organization in Europe in 1988. The unit started off with one person—me—and ended up with around 600 staff, if you count temporary people. So you could say that I built Microsoft's localization facility in Europe. After that I did some consulting work for various people in Ireland, including Oracle. Basically, they asked me to come in and have a look at their localization activity and report to the Oracle product development management committee on what I thought should be done. And then they said, "since you're so good at recommending what we should do, why don't you join and make it happen."

LISA: A consultant's nightmare!

Michael O'Callaghan: Absolutely, and I put up a good fight for a time but then I finally joined. At Oracle, my brief has basically been to implement my own recommendations. Previously, there had been a unit in the Netherlands that had been translating files and sending them back to the US, but it wasn't a true localization center in that sense. And they weren't able to get the additional people to grow the organization. So my target was to actually establish a scaleable localization center for the first time.

LISA: What did you learn from your six and a half years at Microsoft, and what changes did you make at Oracle based on this?

Michael O'Callaghan: We found out a lot of things the hard way at Microsoft. And six and a half years later everyone in the industry—or almost everybody—is obviously much wiser. The level of expertise in the industry is that much greater, and there is a lot more focus on international at corporate level. I would say the biggest change is that there are a lot more people around who actually have the experience needed to get up to speed quickly on new products and technologies.

LISA: How much has Oracle grown in Europe during this time, and how much of this growth is attributable to localization?

Michael O'Callaghan: In the third quarter of 1998 alone, Oracle recorded an increase of 37% in revenue in Europe, but it's very difficult to actually put a number on how much is localization-related. I won't attempt to. We effectively have two businesses: one is our systems business which includes the database, tools and e-commerce platforms, and the other is our ERP applications business. The systems products are infrastructure products and enable customers to build the IT infrastructure they need to support their business. This area has grown significantly but it is difficult for me to say how much of the growth is down to localization per se. On the applications side, it is perhaps a little bit easier, because the products are intended for end users and as such are all localized, so growth in the non-English markets could indirectly be attributable to the availability of localized versions.

LISA: Who decides whether to localize and what to localize?

Michael O'Callaghan: It's a combined decision involving the sales organization and the US product development groups. Some things are done on a strategic basis—for example, there are certain products where we say, "yes, we will do those in all languages"—while for others it very much depends on the people in a country saying that there is an opportunity to sell the product if it is localized. Here it is important to take a view over a number of years, rather than just looking at a single version.

Also, if a product is intended solely for systems people, who tend to have English as their operating language, we are less inclined to localize. If, on the other hand, it's an end-user product, we'll do more of it. So, while we localize all our products, we don't do every product to the same extent.

LISA: How do you resolve conflicts when making localization decisions?

Michael O'Callaghan: It comes down to the business case: if the numbers don't stack up, we won't do it unless there are some good non-revenue reasons. If they do stack up then we will localize the product as long as we have the resources and budget with which to do it.

LISA: How is your localization set-up organized?

Michael O'Callaghan: We have a European localization center in Dublin, and in each of the European countries we have language specialists who are responsible for language quality—for the terminology for the country and for the quality of the work produced by our in-country vendors. In Asia, we have organizations in Bangkok, Taiwan, Beijing, Seoul and Tokyo. These are like micro-images of the European operation.

LISA: What about your vendor base?

Michael O'Callaghan: All our translation work is done in country and so we have a very distributed vendor base. We don't bid jobs—we negotiate annual contracts and call off projects against these. We do quarterly forecasts which our vendor partners use to better predict the amount of resources required in any given period.

LISA: Do you do most of your business with multi-language vendors?

Michael O'Callaghan: What we are actually buying is language, and we try to buy the best quality language for each country. Where we can get this for a number of languages from a single vendor then we will go with that. With the newer languages there tend to be more and smaller vendors. We also try to balance our dependence on a vendor and the vendor's dependence on us to ensure there isn't overexposure on either side.

LISA: What role does localization play with your e-business products? Will English be the dominant language, or will you have to make everything available in Chinese and Japanese, or whatever?

Michael O'Callaghan: Oracle's strategy is centered on network computing, which means that business-critical information and applications should be stored centrally, where they can be professionally managed. All our products are designed for this approach—data resides on the data server, the applications reside on an applications server and all that people need in order to run the accounts payable/accounts receivable or whatever is a browser. This is the core philosophy of how we believe computing will evolve. At the end of the day we still have a lot of localization to do: the end user still has to be able to work with the applications, regardless of whether they reside on a desktop or on the application server. But, obviously, from the point of view of updating, it is much easier for an organization to change applications on an applications server than to go around changing them on desktop machines.

LISA: So a Chinese Oracle user will be working in Chinese, for example?

Michael O'Callaghan: It depends on whether they are running a localized application or not. For example, if someone buys a book from Amazon.com, they are actually an Oracle user because the underlying technology is Oracle technology. However, Amazon.com may not present its data in a local language. Our technology is localized and the parts of the underlying technology that are visible to the end user—error messages and the like—could also be in the local language. However, the ultimate decision as to the extent that local content is provided lies with the supplier of the service using the Oracle technology.

LISA: Is your group responsible for the support in such cases?

Michael O'Callaghan: For a localized product, if there is a localization support issue then this will eventually come to us for resolution. However, the technical support organization in each country, which is the first point of contact for our customers, resolves most customer issues.

LISA: Do you advise your customers what to localize when they're building applications with your products?

Michael O'Callaghan: If the customer is a company that wants to roll out something in multiple countries or convert a product to a different language, our consulting organization in the target countries is capable of helping them localize the product for that market. We work with our consultant colleagues from time to time and we make our glossaries available to them if they require them.

LISA: What are the greatest challenges facing you in the shift to this new type of business?

Michael O'Callaghan: Well, there are new technologies involved, and as these evolve there are always challenges. But I don't think there are any different challenges per se in this type of localization.

LISA: Do you feel your department is getting the kind of support it requires from development and the other departments you collaborate with in terms of hand-off, design, and other internationalization issues?

Michael O'Callaghan: Overall I would say yes. However, there are differences—the more mature products tend to be further along the localization learning curve, while the newer technologies tend to have more issues.

LISA: How do you prevent people reinventing the wheel and making the same old localization mistakes with new technologies?

Michael O'Callaghan: By having the right infrastructure in place, the right processes, the right information. It comes down to training and education, and reiterating the localization message. I would say it depends more on the power of the organization to change than on the products themselves or on any technical issues that may be involved in the new technologies.

LISA: How does Oracle manage its training process?

Michael O'Callaghan: As a localization group, we have a number of people based in the US whose job it is to work with the product divisions, to ensure that translation and localization issues are being addressed. We also have an internationalization group whose role is to help design an internationalization test suite that product groups can use to verify their product for internationalization compliance. At the end of the day, what it comes down to is having a number of people who know what the issues are, and having them get around and talk to as many development managers and developers as possible. It's an ongoing process.

LISA: Would you say that localization is viewed strategically at Oracle?

Michael O'Callaghan: I would say it is used strategically by Oracle. For most products it is extremely strategic, for some products it may not be that critical. It depends on the focus. But overall it is a key part of Oracle's strategy.

LISA: How do you see language technology helping localization? And what changes do you expect to see here within Oracle and within the technology in the next five years?

Michael O'Callaghan: I think we will see a great improvement in the ability to easily reuse translations and I think more and more people will be using machine translation. I also think that there will be a structure for integrating machine translation and translation memories in an automated and cost effective way. Basically, people will use a translation memory to scrub out all the stuff already translated, and then put the rest through a machine translation system and then post-edit it. This will change the role of the translation vendor. This is just a personal view by the way, and I could be totally wrong.

LISA: So you're predicting a swing back to an in-house repository?

Michael O'Callaghan: I don't know whether the repository will be in-house or not, but the process needs to be highly automated. It is possible that some of the vendor companies may actually start to create large-scale industry-specific repositories. For Oracle and most of the software publishers, our core businesses is not localization—it's a strategic requirement to enable us to deliver our products but if there's somebody out there who can manage this repository and do a super job and support the strategic goals of the company, that's fine.

LISA: So you would be open to a totally outsourced solution, if the right technology and partner could be found?

Michael O'Callaghan: The actual process of converting the language is not the only issue to be addressed, but if the right technology were available and the partner could be found we would consider it. Right now it doesn't arise.

LISA: One of the biggest problems in creating this sort of solution is to get the publishers' internal organization right. How are you tackling this at Oracle?

Michael O'Callaghan: One of the things that we are looking at is trying to build some kind of workflow processes. We are all familiar with the situation where people work really hard, do an absolutely fantastic job to very tight deadlines and then omit a file or something right at the end. We resolve all the major problems, do the impossible and then trip up over a simple thing. We want to eliminate this possibility by building an application which will control the process as well as the components in the product itself. The creation of this type of solution is Oracle's core business, so we can call on expertise in-house to help deliver it

LISA: It's also a question of departmental responsibilities. Who owns the repository, who is responsible for updating it, who ensures that the information flows properly across departmental boundaries?

Michael O'Callaghan: I think you have to design an end-to-end process. What we need is to design a product that is capable of being sold in the various target markets around the world, look at the product delivery process from beginning to end and decide on responsibilities and what is the most efficient process and information flow to do that. If you start off from the question of which departments have historically done what, you'll never get to an optimum solution.

LISA: How satisfied are you with your current tools?

Michael O'Callaghan: The tools are never good enough, but we have made huge improvements over the past year. We're just coming out with a new suite of internal tools for some of the software work we need to do, and we also use some of the commercial translation memory tools for the help systems. We are not investing on a large scale in machine translation at the moment, but we are actively looking at how we might use the technology.

LISA: Do you collaborate on a strategic alliance or partnership basis with any of your tools manufacturers?

Michael O'Callaghan: Yes, we're trying to look at designing a better process rather than a specific tool, so we're looking at where tools slot in and where they do not provide the support they should. We've done some co-operative work with a number of companies where we've suggested improvements to off the shelf tools and helped implement them.

LISA: Do you recommend to vendors that certain tools be used for certain tasks?

Michael O'Callaghan: Yes, we tell our core vendors that we expect them to use the tools that we have been developing in-house as well as nominated third-party tools.

LISA: What are the biggest challenges facing your unit at the moment?

Michael O'Callaghan: We'd like to make localization easier. Ten years ago in this business virtually everything was done manually. Today we have better tools and increased automation but the products have gotten bigger and more complex, the localization deadlines are tighter and the quality expectation is higher, so it's still a hugely labor-intensive and stressful process for all concerned. We want to make it less stressful by adding more technology rather than more bodies. Having more stressed people around is not the answer! We don't want to eliminate the people, just make the process smoother and make it easier for the people we have to do their jobs.

LISA: What about education? Are you getting the people you need or are you investing in in-house training?

Michael O'Callaghan: Both. I believe we have great people on board here in Dublin and around the world and are continuing to attract the caliber of person we need. We also make a concerted effort to train people in-house, although if you talked to the people they would probably say we should do more. Our training focuses on developing people skills—management skills, review and appraisal training—as well as technical training to do the job.

LISA: In general, are your service partners meeting Oracle's requirements?

Michael O'Callaghan: Yes, because if they didn't, we wouldn't be able to ship the product. But I have to say that there are things that need to improve. For example, we are working with all our partners to get to a situation where we don't have to check their work because they have in-house systems in place to guarantee the quality of what they deliver. We are trying to change the approach over time so that they have a proactive approach to QA which builds quality into the translation process rather than the current situation where we all check for errors at the end of the cycle.

LISA: Are there areas which you feel vendors should take total responsibility for?

Michael O'Callaghan: Yes, I feel they should be totally responsible for their part of the process. We are asking them to translate something and we would like to be able to rely on them to deliver it 100% correctly. This is probably some way off yet, though. Equally, we want them to be able to rely on us to do our part without screwing it up. It works both ways.

LISA: Are you measuring how satisfied you are with your tools?

Michael O'Callaghan: Our in-house tools people have to determine the cost benefit of the tools that they're developing. It's fine having a fantastic piece of technology but if it doesn't make the process quicker or improve the overall quality, then what's the point of introducing it?

LISA: What skills will be needed in the localization industry in the future, and what do you see as the biggest challenges in this context?

Michael O'Callaghan: This is a very difficult one to answer. I think a couple of things are going to change. We're going to continue to have the traditional type of products and consequently the skills will still be needed to localize these. In addition there will be a requirement to provide Web sites and online information systems in many languages. We need the capability to do that very quickly. If you're running a business and selling in ten or twenty different countries and you take a decision to change pricing or do a special offer, you probably need to be able to make changes in all languages simultaneously. There are companies trying to achieve this today, but nobody can do it yet. On a Web site with a lot of information you can't simply update the English and immediately change all twenty other languages. This will probably be a common requirement in the future.

In terms of the industry itself, I think the emphasis will change from doing a lot of things manually to doing them in a more automated fashion. Therefore the skills mix will change and there will be a bigger requirement for people who can develop the necessary automation. Everyone is trying to automate more because the volumes are getting bigger, there are more products, and as international business becomes more important you have to do more languages quickly and to a higher quality. So on the translation side, the expertise will shift from the ability to translate words towards developing systems to support the translation process.

On the vendor side, the margins issue will affect investment and the future business model. If you can get 5% in the localization industry and 20% somewhere else, why should you invest in a 5% business? I anticipate a strong push from investors for increased profitability in the consolidated localization vendors. Since increasing the charges to the publishers is not really an option, margins will have to be increased by other means. I believe that this will be one of the major challenges facing the supply side over the next couple of years.

LISA: Thank you very much.