Sunday, August 29, 2004

Outsourcing 101: Introduction

Mrs. Smith, a 70-year old retired grandmother from Texas, cannot turn on the personal computer she just bought. She gets frustrated and calls up the support line from the computer company. Joe, with a pleasant Texas drawl, responds on the other end, listens to her story, and asks a few questions. Finally, the problem is identified: she's been pushing the computer logo instead of the power button.

In today's on-demand business environment, product support services like the one Mrs. Smith called are a necessity. They are essential because the products have become more complex, and because customers have come to expect them as a standard service. Failing that, customers could just simply switch vendors.

Let's take this story from the other end, from Joe's side. Is Joe from Texas? He might be, because of his accent. But Mrs. Smith doesn't know that for sure, and neither do we. Then again, Mrs. Smith doesn't care, because to her, Joe is really just a voice.

For all we know, Joe could be actually be Jyothi from Bangalore, an Indian who affects a Texan drawl. Or he could be Jonathan from Makati, trained to do the same. Or he might even be Joselito from Bacong. What's going on here?

Our hypothetical Joe -- whether it's Jyothi, Jonathan, or Joselito -- is actually working in a call center. Trained both in call-handling and in the specifics of a product, they field calls from the many different Mr. and Mrs. Smiths around the world. Why go outside the US? Because it might cost the computer company $15 per hour to pay a real Joe from Texas; but it would only cost them $5 per hour to pay a Joselito from Bacong.

Call centers are but the tip in the massive iceberg called business process outsourcing. Companies outsource because they're looking for cheaper and more efficient ways to operate, especially in areas of their business for which they do not or cannot excel in, and so they hire another company to do it for them.

What are the other areas beyond call centers? Here are a few other examples:

1) Medical transcription. Doctors in the US are required by law to maintain written records of all diagnoses and prescriptions to patients, but they are typically too busy to do it themselves. So they dictate their findings into a recorder, and have someone else transcribe it for them. That transcription can be outsourced.

2) Payroll processing. Companies worldwide must compute for the payroll, taxes, exemptions, and benefits for their employees. While its true that they must follow the business regulations of the countries they operate in, it also just means that the accountant must be familiar with the regulations. That payroll processing can be outsourced.

3) Animation. Movie and advertising companies usually write their own stories and scripts for their animated features. But creating the actual animation is tedious, even with computers. More than that artistic skills are required, which may be in limited supply in the country where they operate. Solution? Outsource the animation.

In fact, think of any service that you avail of that doesn't require a face-to-face interaction, which you can transact via the computer or via the phone. That service is a good candidate for outsourcing. In reality, it may already be outsourced!

So what does all this have to do with Dumaguete? Why, nothing at all, if you think that all we can ever be is a university town and retirement home. But if you think we can be more than that, if you think that Joselito from Bacong deserves that better and higher-paying job (without having to leave Bacong), then it means everything.

Sunday, August 22, 2004

A Meeting with Investors

A Rational Technology repost.

Over the past year, Injong Fortunato, Dean Sinco, and Veneeth Iyengar have thrown upon themselves the challenge of promoting Dumaguete as an investment destination for information technology and outsourced services. It's always a pleasure to join these activities, because it's an opportunity to meet new people and listen to new ideas. Our meeting with the Board of Investments in Manila last August 18 was no exception.

What made this meeting different from the ones that came before it was the powerhouse cast in attendance. Prominent personalities represented the Dumaguete entourage: Fred Dael, Mariant Escano-Villegas, Mike Romero (!!!please double-check with Injong to make sure I didn't leave anyone out!!!). Likewise, the representatives from the investment community came from outsourcing organizations, chambers of commerce, and multinational companies. Things are starting to get serious.

The Dumaguete presentation was as I had seen it many times before, though refined somewhat in many areas from encounters with past audiences. Flashed on screen were our aspirations to become hubs for software development, medical transcription, and business process outsourcing. Augmenting this presentation were the actual figures of the cost of doing business in Dumaguete.

Globe Telecom also made a pitch, revealing a little-known fact: the network bandwidth capacity going into and out of Dumaguete are actually quite substantial, rivalling that of Cebu City. Much of this was established during Fred Dael's tenure as CEO of Islacom, and the infrastructure remains to this day. At last count, the capacity stood at 320 gigabits per second.

How much is 320 gigabits per second? Assuming that a dial-up Internet connection goes at 32 kilobits per second, a pipe that thick will allow for 10,000,000 dial-up Internet connections to go through at one time! This is highly desirable for companies that want to set up shop in Dumaguete.

On the flip side, though, I felt that our potential investors posed tougher questions than before. For one thing, the call center people felt that the population of Dumaguete was too small to sustain a call center labor pool, based on their numbers from Manila and Cebu. There were questions, too, about the transient population, the high cost of power, and other supporting infrastructure. This is a marked contrast from our first meetings which were mostly congenial and encouraging.

Far from being hostile, though, I consider this a positive sign. The first few times, they'll probably smile at you, perhaps a little condescendingly; if you're not serious, you'll go away with pleasant memories of how friendly they were. But now that we've persisted several times, they know we mean business. Now they're beginning to treat us seriously, and the serious questions are starting to come in.

There's still a lot of work to be done, not just for the Dumaguete Trio, but for the city as a whole. But we should all keep at it. We're starting to round a corner, and it would be a shame to let up now.

Sunday, August 15, 2004

Data Mining

A Rational Technology repost.

The beer-and-diapers story is one of those urban legends of information technology. The story goes thus: convenience store chain 7-11 ran an intensive study of their customers' buying pattern over a period of time and found an unusual correlation between the sales of beer and the sales of diapers.

To test their theory, they placed their diaper stocks next to their liquor section. The sales of diapers increased dramatically. Why? Husbands watching Monday night football with their friends would invariably go to the convenience store for beer. Wives would remind them: "Honey, since you're dropping by the store anyway, could you pick up a box of diapers for Junior?"

The story has been recycled many times and I can't say for sure whether it's true or not. But it does illustrate one of the reasons why companies collect and study data. They're looking for those patterns which can give them stategic advantages. This particular field is called data mining. Like prospectors mining for gold, information management specialists also sift through data looking for those little nuggets.

Data mining involves examination of variables to determine the relationships that exist between them. It's easy enough when you're only dealing with a limited set. For example, if you had to correlate sales of softdrinks in Dumaguete with the time of year, you might easily see that sales are higher in the summer and possibly conclude that more people are buying because they are hot and thirsty.

Of course, the real world is far more complex. If you take more variables into account you might see more relationships. Higher sales could correspond with higher tourist influx, promotions, busted water pipes, better distribution, town fiestas, birthday parties, elections, etc. Data mining takes into accounts hundreds of variables in order to paint a better picture of the situation.

In so doing, companies can create mathematical models to simulate the effects of any change in the variables. Will softdrink sales improve by developing better distribution? Perhaps the company should sponsor more town fiestas and city events to boost sales? Perhaps the company should partner with travel agencies? The possibilities are endless, and having an accurate model takes out the guesswork.

And that's the reason why companies collect data of all sorts, and correspondingly, the reason why their data storage requirements can hit the terabyte ranges that I've written about in the past two installments of this column.

Think also about the type of computers required to perform these calculations. If you had to search through 128,000 years worth of Metro Post, making computations along the way, how long would it take you?

Sunday, August 08, 2004

Walking statistics

A Rational Technology repost.

A terabyte of data can hold 128,000 years worth of Metro Post, or 200,000 songs, or 200 movies. These are interesting measurements to contextualize the scale of the information but somewhat removed from our immediate experience.

So let's user ourselves as a unit of measurement. How much practical data could we normally associate with a single person? Let's take an average middle-class Juan as an example.

Top of mind, you would have the usual information associated with any typical application form: full name, date of birth, home address, home telephone number, work address, work telephone number, cellphone number, email address, name of spouse, name of father, name of mother, names of children, schools attended.

Extending this, you might also include the social security number, tax information number, bank account number, religion, previous jobs held, annual salary, organizational memberships, criminal records (if any), favorite color, blood type, allergies, known diseases, etc. All in all, this information can fit into maybe two or three pages of a standard curriculum vitae. Estimated size: around 12 kilobytes worth of data.

At this rate, information on the 300,000 residents of Dumaguete would fit in about 4 Gigabytes of data, a little less than a single DVD.

But wait. That can't be all, can it?

Let's extend it further, looking at the typical transactions a single person makes. Credit card transactions, deposits and withdrawals, loans, cellphone bills, grocery purchases, car payments, etc. In a single month, these transactions might amount to an additional six pages of data. Estimated size: around 24 kilobytes per month.

In a single month, the residents of Dumaguete would take up just 8 Gigabytes, or about two DVDs worth. In a year, that's 96 Gigabytes, roughly a decent sized hard disk. If you want to keep ten years worth of data of all transactions, then that's when you start hitting the terabyte range. At this point, a terabyte of data no longer seems so unreasonable or unattainable.

Of course, that opens more questions: what's the value of keeping this much data, if at all? Tune in next week!

Sunday, August 01, 2004

128,000 years of The Metro Post

A Rational Technology repost.

One of the most interesting aspects of my work nowadays is enterprise data storage. Most people would associate this with computer hard disks, and they wouldn't be too far off the mark. The primary difference is in the scale, as I typically work with disk systems that hit the terabyte range.

What's a terabyte? The quick answer would be 1,000 gigabytes, roughly the equivalent of 12 of the largest-sized hard disks that you can buy. Of course, these figures are a bit abstract, even when you compare it to hard disks. So let's take this newspaper you're holding in your hands as an example.

On the average, my column consists of 3000 characters, and that's equivalent to about 3000 bytes or 3 kilobytes. Since I occupy about 1/6th of a page, and the usual Metro Post issue runs at 8 pages, we can therefore calculate that the text of each Metro Post issue runs to about 144,000 characters, or 144 kilobytes.

One hundred forty four kilobytes isn't much. At this rate, the text of ten issues of Metro Post will fit into a standard floppy disk; a year of Metro Post will fit into five floppy disks.

A standard 80 gigabyte hard disk, something you can typically buy from your PC dealer, will hold about 560,000 issues of Metro Post, or roughly 11,000 years worth of weekly publications. A terabyte of data then runs to about 128,000 years worth of Metro Post. Assuming, of course, that we all live that long, the writers haven't run out of ideas, and you haven't gotten tired of reading us.

These figures look a bit mind-boggling. Do we actually need this much data? Admittedly, it's too much Metro Post for anyone to handle, but the reality is we live in a world of very rich media. Beyond words, we deal with pictures, music, and movies. At this level, a terabyte of data starts to look more reasonable.

Your typical high resolution picture, as shot from a digital camera, is about one megabyte in size. A one terabyte hard disk will hold about 1,000,000 pictures. Still a bit much? Let's move a couple of steps higher still.

Your average song, in MP3 format, runs to about 5 megabytes. A one terabyte hard disk will hold around 200,000 songs.

Your DVD disc, containing two hours worth of a movie, holds 5 gigabytes worth of data. A one terabyte hard disk can therefore hold up to 200 movies.

At the level of DVD movies, a terabyte worth of data starts to make sense. And yet, this is only the tip of the iceberg. The reality is: there's still so much data out there that it makes a terabyte look paltry.