Nobody’s Perfect

Blog Administrator | Analyzing Data, Analyzing Data Quality, Data Enrichment, Data Management, Data Quality | , , ,

By Elliot King

People of a certain age may remember the long-time marketing slogan for Ivory soap. Ivory was 99.44 percent pure (pure of what was left unsaid.) While it would be nice if our data could be 99.44 percent accurate–or better yet 100 percent accurate–that goal is simply impossible. Data degrades over time. Telephone numbers, physical and email addresses and so on that were correct today will not be correct tomorrow. Life happens and that changes your data.

So how good is good enough? The answer to that question is not a specific
number or threshold. Instead, data must be good enough to meet the needs of the
data users.

There are two ways to go about achieving that goal. The first option is to
respond quickly when data users complain. This is the most common tactic and
unfortunately both the most risky and the most costly. Faulty data can have a
huge impact on business processes and that impact is generally not for the
better. When those kinds of major hiccups occur, not only the data must be fixed
but the disruption caused by the flawed data must be fixed.

A better strategy is to clearly understand the needs of data users and then work
systematically to meet those needs. Let’s consider a marketing campaign. How
many of the records contain all the information needed to target and communicate
with the intended audience? How many complaints from people who do not wish to
receive your message do you receive? What are the mechanisms you have in place
to address those concerns? Does your marketing database cover your market area
efficiently? Similar sorts of criteria can be developed for other business
processes that rely on data.

Once the needs are established, quality targets can be achieved. From that
point, data quality becomes a real numbers game. How much does it cost to
improve data to the point that it has a significant impact on a business
process? At a certain level, incremental improvements are more costly than the
gains they provide.

At the end of the day, data is good enough when it satisfies the needs of the
users within a reasonable budget. And that is a goal you can achieve and is
close to perfection.