What Is Data Quality?
reality and the data in your system do not agree, you have a data quality problem.
The system says you have 10,000 cases but you actually have 9,850. A customer's
ship-to address is out of date. You have the same piece of information in two
places in two applications or even two systems, but they do not agree.
things can cause data quality problems. Transactions may have been inaccurately
recorded, 1002 units were picked but 1020 were recorded. Maybe the physical
act was flawed, a pick list calls for 1,000, the transaction is recorded as
1,000 but only 900 were picked. When an item is placed in a location and the
location is incorrectly reported, the result is two problems, 1) the location's
inventory is overstated, and 2) another location's inventory is understated.
delays can cause temporary errors. The recording of the transaction is delayed
so that for a period of time, the data does not reflect reality. No harm, unless
a decision is based on the incorrect data.
quality sounds like a motherhood and apple pie issue, of course we want our
data to be right. However, very few enterprises get serious about it. Maybe
that's because the cost of data quality is hidden. That cost can be huge.
What Are The Costs?
Data quality problems cost money. Internally, decisions based upon flawed data results in poor decisions. Fixing the problems caused by data quality takes time and money. Fixing the data itself is also expensive. Today, with extended supply chains and collaboration, the problems and the costs are shared by trading partners as well.
Let's look at some examples. A recent study by the Grocery Manufacturing Association (GMA) and the Food Marketing Institute (FMI) focused on the impact of data quality in the supply chain. The study estimates that:
12 23% of sales and administrative personnel time is spent correcting invoices,
orders, purchase orders, etc.
3 5% of all stock outs, resulting in a loss of sales, are due to data quality
Item catalog errors take on average $60 80 per error to fix and catalogs
average an error rate of over 30%.
If inventory data is wrong, you either lose sales due to an inability to ship or, more often, increase your safety stock to avoid future lost sales. Either revenue is lost or capital is used as a result of data quality problem. When sales orders or invoices are incorrect, time is needed to correct the data quality problem meaning both a manual effort and increases in the order to cash cycle resulting in your funding an increased accounts receivable.
If a data quality problem impacts business partners, the cost may not show up easily. Customers usually do not tell you that they are ordering less or vendors tell you they are charging more due to the problems caused by data quality. If a customer discontinues or cuts back on the relationship, data quality could be a contributor (or the total) problem.
The cost of data quality is real. The costs can be large. Since company financials lack a line for "cost of data quality" it is often ignored and considered a cost of doing business. It is not.
What Can Be Done About It?
First, if management does not acknowledge a data quality problem, it cannot be fixed. Second, unless management sees the data quality problem is creating business problems, it cannot be fixed. In most companies, if a cost is not associated with the problem, it will never be given a priority and will continue forever. Therefore, the first step in addressing a data quality problem is to identify problems, the resulting impacts on the business and the cost of those impacts. This calls for both a broad picture and detailed analysis of specific issues. Management will need specific evidence of customer X being short shipped, ingredient Y being out of stock, and the real numbers associated with both.
Given the right priority, you must first find out where you have a problem and then how the problem was created. A physical inventory may tell you there is an inventory accuracy problem, but it can only be fixed if you determine how it occurred. The entire cycle must be inspected looking for sources of data errors.
Once the how is identified, you must address it. Do physical barriers exist? For example, since you do not have enough racks, warehouse personnel have to estimate certain quantities. Does the basic business process create the errors? For example, you hold all receiving documents at the dock for 24 hours so the foreman can review them, resulting in the data being late. Is training required? For example, have new forklift drivers been trained as well as the older ones? Do you need to implement more error checking procedures? For example, if we ship in full pallets but report in cases, do the numbers make sense?
One of the most frequent and most difficult causes of data quality is culture. If people do not think that data quality is important, it isn't. Since it is not important, people do not worry about the accuracy of a count or their handwriting on a picking document or they guess at the slot number. People will always be people and they pay attention to what's important. Culture is the hardest thing to fix. Culture change can only start at the top. Management has to see the problem as a culture issue and continually demonstrate its importance to all.
company has data quality problems. No one is perfect. The question is, "Do your
data quality problems impact the business enough to demand the problem be addressed?"
Data quality problems do not come to light easily, people do not want to fix
them; they want to ignore them. But these problems are costing you money. Assume
you have data quality problems and continually check on the costs of these problems.
Will the cost of fixing the problems be justified by the elimination of the
cost of problems themselves?
Thompson is a principal of Process ERP Partners. He has over 25 years
experience as an executive in the software industry. Olin has
been called "the Father of Process ERP." He is a frequent author and an award-winning
speaker on topics of gaining value from ERP, SCP, e-commerce and the impact
of technology on industry. He can be reached at Olin@ProcessERP.com.