Page 56 -

Timely Q1-6 What Are Necessary Data Characteristics? 55
Good information requires that data be timely—available in time for its intended use. A monthly
report that arrives 6 weeks late is most likely useless. The data arrives long after the decisions
have been made that needed your information. An information system that sends you a poor
customer credit report after you have shipped the goods is unhelpful and frustrating. Notice
that timeliness can be measured against a calendar (6 weeks late) or against events (before
we ship).
When you participate in the development of an IS, timeliness will be part of the requirements
you specify. You need to give appropriate and realistic timeliness needs. In some cases, developing
systems that provide data in near real time is much more difficult and expensive than producing
data a few hours later. If you can get by with data that is a few hours old, say so during the
requirements specification phase.
Consider an example. Suppose you work in marketing and you need to be able to assess the
effectiveness of new online ad programs. You want an information system that not only will deliver
ads over the Web but that also will enable you to determine how frequently customers click on
those ads. Determining click ratios in near real time will be very expensive; saving the data in a
batch and processing it some hours later will be much easier and cheaper. If you can live with data
that is a day or two old, the system will be easier and cheaper to implement.

Relevant

Data should be relevant both to the context and to the subject. Considering context, you, the
CEO, need data that is summarized to an appropriate level for your job. A list of the hourly
wage of every employee in the company is unlikely to be useful. More likely, you need average
wage information by department or division. A list of all employee wages is irrelevant in your
context.
Data should also be relevant to the subject at hand. If you want data about short-term inter-
est rates for a possible line of credit, then a report that shows 15-year mortgage interest rates is
irrelevant. Similarly, a report that buries the data you need in pages and pages of results is also
irrelevant to your purposes.

Just Barely Sufficient

Data needs to be sufficient for the purpose for which it is generated, but just barely so. We are
inundated with data; one of the critical decisions that each of us has to make each day is what
data to ignore. The higher you rise into management, the more data you will be given, and because
there is only so much time, the more data you will need to ignore. So, data should be sufficient, but
just barely.

Worth Its Cost

Data is not free. There are costs for developing an information system, costs of operating and main-
taining that system, and costs of your time and salary for reading and processing the data the
system produces. For data to be worth its cost, an appropriate relationship must exist between the
cost of data and its value.
Consider an example. What is the value of a daily report of the names of the occupants
of a full graveyard? Zero, unless grave robbery is a problem for the cemetery. The report is
not worth the time required to read it. It is easy to see the importance of economics for this
silly example. It will be more difficult, however, when someone proposes new technology to
you. You need to be ready to ask, “What’s the value of the information I can conceive from
this data?” “What is the cost?” “Is there an appropriate relationship between value and cost?”
Information systems should be subject to the same financial analyses to which other assets
are subjected.