by Marjorie M.K. Hlava
As more and more people look into the computerization of their database, whether the internal files are bibliographic or full text, they quickly realize that building a database is highly labor-intensive.
The task of either keyboarding or scanning all of that data and getting it into an online system can often be a large one. Since labor rates in the United States are high compared to the rest of the world, people are looking increasingly outside the country for the keyboarding of their textual information in order to lower costs.
Although banking information and airline ticket data historically have been keyboarded offshore (using a 10-key approach), the idea of using full keyboard digitization of information is relatively new. Only a few pioneers in the information industry have used offshore keyboarders in the past, and the whole concept of using an offshore keyboarder is a new one to our marketplace. As in all other segments of the worldwide import/export economy, there are both honest and solid business people and those who use less ethical means to get and produce a product. Because the offshore keyboarding business is no different, this article strives to outline some of the things to look for, how to run a test of the data, and how to monitor it on an ongoing basis.
There are a great number of offshore keyboarders, and they are usually represented in the U.S. by a marketing firm. Some of these marketing firms actually have a part ownership, but many of them are independent and work essentially with a subcontracting arrangement to the offshore keyboarding facility. Quite a few have a clause in their contract saying that you may never contact the offshore keyboarding facility directly yourself. Depending on the size of your job, this may or may not be a disadvantage.
Who are the offshore keyboarders?
Offshore keyboarders are located in many countries around the world. Some are English speaking and some are not. Examples are International Computer Resources with offices in Stanford, Connecticut, and a subcontracting arrangement with an outfit in Seoul, Korea; Offshore Keyboarders in Tarrytown, New York, with an offshore keyboarding relationship in mainland China; Saztec International with a subcontracting arrangement primarily to a facility in the Philippines; Access Innovations, Inc. with a subcontracting facility in Mexico City. There are also groups with large facilities in Ireland, Scotland, England, Barbados, Jamaica, and India.
Indeed, most of the facilities are what we would call in the U.S. traditional sweatshops. The employees are tied to a certain number of keystrokes per hour with deductions for errors. The keyboarding is often done by people who do not speak or read English; most of the people are trained as they would be trained in a 10-key numeric operation, i.e., by character recognition.
The character recognition basis has both strengths and weaknesses. The strengths are that the data is keyed much more quickly and is keyed exactly as it appears on the page. The drawback is that people who are looking at it are not reading the text and if you have any little marking like “Turn the page over, please” or “More on the back” or “See above,” those items will be faithfully keyed in along with the rest of the information. Any editorial notes will be correctly added in the appropriate place but you will not get the expected reaction to the editorial note.
Therefore, data has to be more carefully prepared for shipping to an offshore facility. Handwritten material is often misinterpreted if an “n” looks like a “u,” or an “a” and “u” are similar in appearance, or if an “i” is made like an “e” or an “1.” Generally an offshore keyboarder will charge a higher price for handwritten data than for completely typed text.
Since offshore keyboarders are trained in character recognition, how do they maintain their accuracy rate?
Well, accuracy rates vary considerably with the typist, as you know from work in your own office. The same is true with keyboarders anywhere in the world. In order to maintain or to guarantee an accuracy rate, most keyboarders will offer a verified, or double keyboarded, price for the data.
Verification can happen in a number of ways. The first way and the most accurate is to have people keyboard the data twice in two separate locations and then run the data against itself to see if there is a character-for-character, space-for-space match of the data. Any time characters are not an exact match, a little bell rings or the machine stops. Somebody who is familiar with the data–preferably someone who speaks English–will then check the data and make the appropriate correction.
Any time characters are not an exact match, a little bell rings.
The second way for verification is “overkeying.” Over keying is when the data is keyboarded and then the file and data are passed onto the next keyboarder. That person then keys over the top of the original data. Any time there is a mismatch, a little bell or signal is given and the person rekeying checks to see what the problem is. This method is a little less reliable than double keying, but still has a very good accuracy rate. The drawback to it is that if you are dealing with handwritten data somebody is going to trust a prior interpretation of that character and just go on.
Yet another kind of verification is to run the data against a spelling dictionary to make sure that every word in the data matches the words in the dictionary. When there is not a match, a string of characters may printout to indicate that this word does not match the dictionary. Of course, proper names, corporate names, and jargon terms do not fit well into dictionary checkers.
The last way to have the data conform to high accuracy standards is to have it proofread. Proofreading works very well for content information and making sure that formats are correct. It is difficult to find proofreaders among offshore keyboarders because they read faster in their own language than by character recognition. Most often, this method is not employed by them.
There are a number of ways to charge for keyboarded data. One of them is character lines. In the United States we often assume that a line is 80 characters (or 80 columns). That is not necessarily true with offshore keyboarders. The length of the line can vary considerably from only a few characters to 132 characters and sometimes longer. If your data is set in a format that requires a line of only four characters for the date field and possibly a two-character tag, for this seven-character line you maybe charged for a 40- or even 80-character line by the offshore keyboarder.
Another way the keyboarder may charge is per 1,000 characters. This would mean they count every keystroke, not just a character but also every space occupied on a line whether it is a character or not. You will need to define whether a capital letter is charged as two keystrokes for the shift and one for the letter itself or as one. Those details need to be checked whether you are paying for column space in a line or whether you are paying for actual keystrokes. Other people may charge for characters only.
All of these charges may or may not include verification. You should request to see in writing that the verification is included.
The best way to check for the actual cost of the data that you are going to have keyed is to have a test sample run. It should be large enough to give you a feel for what the error count might be like. If the sample is more than 30 units, it is reasonable for the keyboarder to ask you to pay for the production of that data; that is, anything more than about 10,000 characters should be paid for by the requesting organization.
When you get the data back, count the characters or estimate them yourself and see what you will be paying for. Getting a bill for the data will actually help you to know what your larger job is going to cost.
Many keyboarders add hidden costs and you need to know what these might be. In addition to the original keying and verification, if they run it through any kind of format or error-checking routines or against an authority list, there will probably be an extra charge.
The production of the tape is often charged in two parts: (1) for the procedure of making the tape, and (2) for the physical tape itself. If you provide tapes or return them, this charge may be waived.
Postage and handling and other kinds of special services may also be charged. This is particularly critical if you are dealing with an offshore keyboarder because the costs of shipping data can be very expensive.
It is also interesting to find out if someone from the offshore keyboarding office is going to walk your tape through Customs. If it is going to go through a regular Customs check, that means it may be x-rayed (and ruined) or checked in some other fashion. Those organizations that use a courier service or have a courier service of their own usually take less time to clear Customs.
You can normally figure turnaround time yourself based on how many keystrokes per hour you think somebody might be able to do on your data. If it is handwritten and is fairly legible, you can estimate that they can keyboard 5000 to 8000 keystrokes an hour. If the data is type script and there isn’t any field formatting to the data, they can probably key in excess of 14,000 keystrokes an hour. If the data needs heavy fielding, the rate will be somewhere in between if it is based on typewritten copy.
Of course, verification of data will have that same speed. Someone typing from handwritten copy will do 5,000/hour on the first keyboarding, 5,000/hour on the second. That means the average amount of data produced per hour will be 2,500 keystrokes in order to get one batch of verified data.
We recommend that you get to know as much as you can about the offshore keyboarding facility and its constraints. If possible, visit the facility. Even through it may seem a bit of a boondoggle to your superiors to look at a facility in far away places with strange-sounding names, it is valid to see if these organizations are really all they say they are.
As in any other business arrangement, be sure that you feel comfortable with the person representing you. If you are concerned that you might have trouble dealing with them without everything in writing, then do not deal with them at all.
Do remember that offshore keyboarders must try to work with significant amounts of data. That means that a job of two million characters is probably not of interest to them. The job should be larger than that for them to consider staffing up and scheduling production.
I am sure that I haven’t covered everything, but I think this will give you a general idea of what you need to know before considering an offshore keyboarding contract. The small risk makes it worth investigating.
We would be interested in learning about your experiences. If you find an excellent keyboarder offshore we would be delighted to hear about your dealings with them. The same is true of one that you think is not particularly reputable — one you would not deal with again.
Best of Luck!