Working together for the quality of data

I would like to begin this post by asking a question: How many contact telephone numbers do you think your administration has? And I will give you the following options from which to choose:

• Many, almost one per taxpayer
• Many, more than one per taxpayer
• I assume that many, but actually don’t know and I do not know whether someone knows.

A personal experience: For several months I had been receiving calls in my mobile phone from a bank’s collection center looking for a Mrs. Martínez. I assume the lady had a significant pending debt with the bank in question, since only thus can one explain the fact that the calls were extended for several months, initially being made during weekdays and office hours, but subsequently extended to evenings and early morning hours as well as weekends and holidays.  It was useless to confirm every time that this was not the lady’s number and that I did not know her. A week later, another voice but with the same accent, again tried my number as point of contact.

The issue became annoying. It was only after talking to the supervisor’s supervisor that I managed to stop the calls.  Nevertheless, the nuisance made me ask myself what had happened, to which I found several alternatives. That the phone number given me had been previously assigned to said lady by the telephone company; that the lady had deliberately given a wrong number in order not to be located whenever she had a debt with the bank. Or perhaps the lady reported the correct telephone number but the person who entered it in the system might have made a mistake; maybe a programmer made a mistake when trying to update all the telephone numbers by including an additional prefix and the result was my number. Maybe it was never reported by Mrs. Martínez, but rather by someone the lady had given as reference and on listening that a bank was trying to locate her, invented a number where she could not be found, which coincided with mine,  and which, as the procedure indicates, was added as one of the possible contact numbers of the debtor. I finally told myself that I was lucky; it could have been worst. The lady might owe money to more than one bank; or the number could have corresponded, as it happened to someone I know; a dry cleaner’s where apparently many people used to call to find out whether their shirts were ready.

Indeed there is a problem with the quality of data, to which I believe we do not attribute importance. I decided to take a look at what happens to that telephone number in our records at CIAT. In one case, in a record of persons who are probably not interested in being contacted, only 66% of the numbers could be possible.  I do not mean to say they are correct, except that the lengths and formats of the other 34% are not compatible with a real telephone number. For example, number 0000000 in some countries will connect us to the operator, because of the first 0, and in others, with nothing. I looked for another record, a smaller one; certainly smaller, wherein the persons might be interested in being contacted. As I had expected, the numbers were better: 91% of the records had numbers that could be viable, although not all could be directly dialed through a computer.

I immediately wondered what could be happening in the administrations. My speculation regarding the composition of the number was that it would be better, since on dealing with local numbers, one could probably reinforce validation rules in capturing the fields. Upon examining some real statistics of an administration to which I had access some time ago, the first surprise was the significant number of empty fields. Then the evidently malformed telephone numbers amounted to 10% of the true records. In sum, more than half of the taxpayers did not have viable registered telephone numbers. With respect to the other half, we know that the number may be viable; but if it actually exists and above all, if it corresponds to the taxpayer, we can say less.

My next questions were if that taxpayer record looks more like our record in that people are really interested in being contacted, or that in which said interest might not be as great and also, what would happen with the addresses, the legal representatives or partners, to mention things related to the record, or those that come in the returns, the substitutive ones, debts and payments, just to mention some of a transactional nature.

Certainly the computerized systems, the electronic returns and payments, even more so, the  MDM (1) type tools contributed to said quality, but, as perhaps some officials working in the collection area might tell us, there may always be problems. Sometimes the problem is in the contact number; on other occasions, hopefully the less, it is in whether the debt actually exists. In some country, all of the payments reported by the banks which could not be applied to anything were affectionately called “the pot”. And that pot only continued to increase.

Improving the quality of data will not be an easy task. It will be a long term process wherein one must identify the sources of the problems and correct them, identify and eliminate existing errors; assign responsibilities, include tools and continue in a permanent cycle for achieving gradual improvements.

I now ask myself: who chose the third option of the initial question and how many administrations have a process aimed at improving the quality of data?

Good luck

479 total views, 3 views today