Return to Decision Line Home Page
Return to DSI Home Page


INFORMATION TECHNOLOGY

Lance B. Eliot, Feature Editor, Eliot & Associates

Year 2000 Outsourcing: Trust but Verify

by Lance Eliot, Feature Editor

The Year 2000 is getting closer by the day, the hour, and the minute. Many firms are just now scrambling to deal with their Year 2000 problem (also known as Y2K), and have decided to outsource most of their Year 2000 activity, including outsourcing the typical range of Year 2000 tasks from initial inventory and impact analysis to actual code modification and testing.

Outsourcing the Year 2000 activity does offer a number of attractive benefits. For example, by using outside specialists instead of in-house staff, the staff is able to continue focusing on day-to-day operations issues and can off-load Year 2000 headaches to the outsourcer. Also, investments in specialized Year 2000 tools can be avoided by using outsourcers that make such investments and then spread their costs across an entire customer base.

Unfortunately, Year 2000 outsourcing can introduce a number of disadvantages that some firms have recently discovered to their dismay. Hopeful of putting their Year 2000 fix into an autopilot mode, some firms have handed over their entire Year 2000 activity to outsourcersþand then allowed the outsourcers to forge ahead without any substantial monitoring. Such a Y2K autopilot approach could lead to a bumpy ride, possibly even producing a devastating crash in your Year 2000 fix efforts.

Safeguard Needed

As an expert in monitoring Year 2000 outsourcing arrangements, I have a number of clients that rely on me to serve as their Y2K safeguard. My motto continues to be: Trust but verify. I aid companies in finding ways to monitor the Year 2000 outsourcer so they can be assured that the outsourcer will do the job as requested.

In my experience, even the most notable Year 2000 outsourcers are susceptible to many of the failings that I will describe in the next section of this article. With the tremendous surge in demand for Y2K outsourcing, vendors are jumping into client engagements overnight and spreading their talent pool dangerously thin.

Firms contracting for Y2K services are often unaware of the pitfalls associated with Year 2000 outsourcing since they are not Year 2000 experts themselves. And, firms are understandably in a rush to make progress on their Year 2000 problem and may inadvertently overlook critical checks-and-balance provisions in their eagerness to push ahead on the Y2K issue.

Common Y2K Outsourcing Problems

Here are some of the more common Year 2000 outsourcing problems that have struck most of the firms that I've been dealing with. The list isn't complete, merely a representative sampling.

Because of the inherent nature of the Year 2000 issue, firms often cannot realize on-their-own that they are experiencing Y2K outsourcing problems until the Y2K effort is nearing completion. Thus, I highly recommend that firms establish either their own in-house Y2K expert to oversee the outsourcer, and/or use an outside Y2K monitoring consultant to help oversee the chosen Year 2000 outsourcer. Do so as soon as the Year 2000 effort begins (usually starting with the inventory and impact analysis).

Generally, the sooner that Y2K outsourcing problems can be identified and corrected, the less impact that the outsourcing problem has on the overall Y2K effort. Trust, but verify, and do so right away.

Now then, here are some pitfalls to be watchful of.

Use of Generic Labor That Lacks True Y2K Training

Date handling is simple, right? Any competent programmer should be able to do it, right? Wrong. There are specific techniques for identifying date-related problems in code and data. Many outsourcers are hiring generic programmers who are versed in particular languages and systems (e.g., mainframe COBOL, midrange RPG), and then throwing them into Y2K engagements.

Without specific training on Year 2000 programming and analysis techniques, the generic labor is likely to miss Y2K problems, will work inefficiently trying to find Y2K problems, and will produce an overall result that appears fine but is riddled with omissions. Make sure that the Y2K technicians assigned to your account are properly trained and experienced in Y2K matters.

Use of Labor That Includes Underperformers

Anyone who has ever managed programmers knows that a good programmer is a magnitude better than an average programmer. Many outsourcers are hiring any programmers that they can find to help fill the huge demand for Y2K services. Unfortunately, some of these newly minted Y2K programmers are actually the rejects who performed poorly elsewhere and have floated in the marketplace looking for a new home. Even if trained in Y2K techniques, the underperformers will plod along, taking longer than necessary, and being sloppy in quality of work performed. Don't let your firm be their next resting homeþmake sure that each person assigned to your account is worthy.

No Traceability

Can the vendor tell you the W's of their work? Who looked at a particular program or module? When was it examined? Where did they look in the program? And so on. The vendor should be keeping a detailed log of their efforts. I am not referring to a billing log for hour tracking, but instead a log that provides Y2K traceability in case questions arise later about the analysis and work performed. A Y2K vendor should have a database that they use to track their efforts and report upon the effort accomplished.

Blind Use of Y2K Tools

Some Y2K vendors make use of specialized Y2K tools to perform various aspects of the inventory, testing, coding changes, etc. I am a proponent of the use of Y2K tools and view them as a good way to speed-up the effort and reduce omissions. But, some vendors also rely on the Y2K tools without a complete understanding of the limits of the tools. Lingering in your code and data are date problems that no existing Y2K tool can spot. If machine scanning is used excessively, you will be misled into believing some Y2K problems do not exist that may indeed exist. Find out which tools are being used and whether the vendor really knows how to use them.

Paying Twice for the Same Work

Suppose that your chosen vendor does your inventory and impact analysis but skimpily records their work (i.e., traceability). Later, when a full-scale analysis is conducted, the vendor will often revisit the same areas as if for the first time! In other words, because of their own inability to trace and document their effort, you pay twice. Make sure the vendor is fully tracing AND documenting what they do.

Scorched Earth Policy

Suppose you grant your vendor the first piece of the Y2K activity, namely inventory and impact analysis, and then open the remaining activities to rebidding. If the vendor does not get chosen for the subsequent steps, they often take a scorched earth policy and delete relevant and helpful records that could be used by the incoming vendor. When contracting for the Y2K outsourcing, make sure that access and ownership to all such relevant records are retained by you.

The Y2K Hollow Survey

Here's a clever labor saving approach that some Y2K vendors adopt. They hand a Y2K survey to your programmers and essentially say ``do my work for me'' to the staff. The vendor shows up a few weeks later to collect the surveys, and the staff hurriedly fills them in to get the surveys completed. The vendor treats the completed surveys like gospel, even though the in-house staff didn't seriously complete them, and the surveys are presented to management without any sensible auditing of their results. It's a hollow result based upon the old Garbage In Garbage Out adage. Make sure a survey approach is workable if you choose to allow it.

Sampling Means Fuzziness

In order to reduce costs or speed-up efforts, some vendors use sampling as a means to examine code and data. They might look at 10% of a particular module and then extrapolate to make conclusions about the remaining 90%. You need to know if such a sampling approach is being used since the results are no longer absolutes þ the results are probabilistic. And, you need to know how the sampling was conducted, by whom, criteria used, etc.

Conclusion

Many clients ask me if there is a magic bullet that will deal with their Year 2000 problem. Sorry, no such tool exists. Likewise, handing over your entire Year 2000 fix to a vendor is not a magic bullet. The best intentioned vendor can still make mistakes when trying to help your firm deal with your Year 2000 problem. I urge you to take a Trust But Verify approach if you outsource your Y2K mess. It's better to be safe than sorry.

Remember that your input is welcomed. If you have projects addressing the information technology area, and you would like to share this with the readers of þInformation Technology,þ please contact me.