Building a successful business requires data, but not just any data: it must be accurate, high-quality data. Acquiring and maintaining a database of good data is no easy feat. For many brands and retailers, quality data can be hard to come by and the costs can add up quickly.
However, it’s important to remember that the cost of acquiring accurate data is far better than the cost of operating with bad data. In fact, bad data can take many forms and have a negative impact on your business far beyond monetary value.
The Causes of Bad Data
We spoke to a number of retail companies about what has led to data-quality issues for their businesses. They identified the following as root causes of bad data:
Poor Communication & Unclear Terminology
Do all your employees communicate with each other? Are your departments siloed or do they share information? Poor communication, or communication that is confusing and riddled with unclear terminology, can easily contribute to poor data quality within an organization.
“Having a clear and concise glossary of definitions for terms is important because it can be different at different businesses,” an apparel retailer told Wiser. “If I say, ‘How many customers have purchased one time in the last six months,’ is it six months from zero or from one month ago?’”
Miscategorized SKUs & Matching Errors
How data is collected, and whether it’s collected properly, are major factors that contribute to poor data quality. For example, the price of a 5.0 oz tube of paint may be compared to a competitors’ 0.5 oz tube of paint.
“We once found that one of our biggest drivers was miscategorized in a major country, and that was really throwing off POS,” a manufacturer explained. “Whenever we saw something that looked out of place, maybe out of category, we would call the retailer and let them know and ask them to correct it.”
Coding Inconsistencies
Does the product on the shelf match the product in your system? The retail industry relies heavily on coding, but if that isn’t accurate, it can throw off all your inventory and sales data.
“If retailers are not coding things correctly, it can cause a bunch of issues,” a consumer products brand noted. “For example, we’ve seen people who abbreviate things, some people put a period, some people don’t. If there’s no consistency in how it’s coded, that creates a lot of issues in data.”
Human Error
Human error, naturally, plays a major role in data quality. The impact of human error is multiplied for any process that is typically automated but requires occasional human intervention, such as if an item doesn’t scan and a store associate must instead enter a code manually to process a purchase.
“Someone may not be classifying something correctly, they may not be capturing something correctly,” the manufacturer emphasized. “The codes might not be written correctly. Those are the two biggest causes of bad data: human error and sampling error.”
The 1-10-100 Rule
What is the cost of poor data quality? Depending on the cause, the effects can be felt in a variety of ways.
This is best illustrated by the 1-10-100 rule, created by Yu Sang Chang and George Labovitz in the early 1990s. The 1-10-100 rule succinctly breaks down the cost of bad data across three core stages: prevention, remediation, and failure.
Prevention
Prevention costs $1 in the 1-10-100 rule. It only costs $1 to validate data and ensure its accuracy at the time of collection. This refers to both time and money: it’s quicker, easier, and cheaper to verify data accuracy immediately than to wait.
Remediation
It costs $10 to fix data-quality issues at a later date. In the 1-10-100 rule, remediation costs 10 times as much as prevention or $10 in this example. Employees will be tasked with identifying bad data, removing it, or replacing it. Your business decisions will be put on hold until accurate data is acquired, or past decisions will be called into question after being based on poor data.
Failure
It costs $100 to do nothing about bad data, or 100 times more than prevention. Your business is a house of cards without systems in place to ensure accuracy. It is only a matter of time before revenue and profits decrease, customers lose trust, employees are frustrated, and turnover increases. All because no effort was placed on preventing bad data from seeping into your organization.
The Costs of Bad Data
Bad data can have a major impact or a minor impact, but no matter what, it will influence your organization.
Unknown Monetary Costs
One of the biggest costs of bad data is financial, but the challenge with data-quality related losses is that they are hard to quantify. Data is used in all facets of your organization, so operating with bad data impacts many people and departments. A decision made a year or more ago may still have ramifications on today’s operations.
“You could be talking just a few thousand dollars, or you could be talking tens of millions of dollars,” A packaged foods company told Wiser. “The cost could be anywhere from nothing to your company goes under, just depending on how much you’ve invested.”
Time & Resources
The remediation stage of the 1-10-100 rule is back. Bad data leads directly to wasted time and resources within your organization in order to fix these issues. You’ll need to spend time validating existing data and remediating errors, which can take days, weeks, or months depending on the amount of data in question.
“If data was wrong and the brand resonated with a completely different group, then you will have just misdirected all that media and all of that marketing, the messaging and the communications, to the wrong group of people with the wrong messages,” according to a beverage brand.
Loss of Credibility
Executive leadership must have total faith in the data presented to them. Bad data weaken that faith, impede decision-making and lead to executives losing confidence in the quality of data in your organization.
“Any time there’s bad data, even if you show It and realize it and go fix it, it creates confidence issues,” the consumer products brand explained. “People don’t trust it as much. In the future, even though you got it right, they won’t trust it and they won’t make decisions based on it.”
Are you worried about bad data? Remember the 1-10-100 rule: it’s far more effective and affordable to identify and fix data-quality issues immediately than to wait. Don’t let problems simmer beneath the surface: take charge and have confidence in your data.
