Organizations rarely regard data entry as a key strategic operation. In spite of the rise in automation, companies across the world rely on manual data entry for procurement. They employ staff to transfer data from requisition orders, purchase orders, and invoices into enterprise resource planning (ERP) systems such as SAP, Oracle, PeopleSoft, Netsuite, and many others.
Transcription and transposition errors are ubiquitous sources of unreliable data for companies that rely on manual data entry. Even the best data entry professionals make significantly more errors than automated systems. Humans are prone to distraction and fatigue, and the most productive user cannot match the pace of the automatic transfer of data between integrated platforms.
Even if data entry is not considered a strategically important operation, procurement certainly is, which is why so many organizations adopt eProcurement and ERP business systems and spend time putting data into them. But it is counterproductive to invest in these business platforms and then rely on error-prone manual data entry.
Data entry is of strategic importance because faulty data is a significant risk.
This point has been demonstrated time and again across multiple industries. Data entry errors in the healthcare industry put lives at risk. Data entry errors in science frequently damage the validity of experiments. Data entry errors in business expose organizations to compliance risks and waste large amounts of money. And, according to The Data Warehouse Institute, data entry errors in procurement, supply chain, and other areas cost businesses over $600 billion each year.
Businesses that eliminate manual data entry in procurement processes with punchout catalogs and order to invoice automation have better data to work with and suffer fewer losses due to data entry errors.
G. Labovitz and Y. Chang developed the 1–10–100 rule to articulate how poor quality control increases waste and costs. It is often applied to data quality.
The first figure (1) applies to the cost of preventing bad data from entering the system in the first place. The second figure (10) conveys how remediating bad data once it has entered a system is an order of magnitude more expensive. The final figure (100) represents the cost of bad data to a business that chooses to do nothing about it.
These figures are general and serve to illustrate a point: it’s far less expensive to implement solutions that prevent bad data from entering a system than it is to fix bad data or to suffer the consequences. Investment in automation is almost always a net benefit to a business that relies on accurate data.
Businesses that opt for manual data input and have no plans for remediation must accept that some proportion of their procurement data is incorrect. The 1–10–100 rule suggests that living with bad data is the most expensive option by a large margin, but where do those costs come from?
The most obvious cause of inflated costs originates with data that is incorrectly entered into ERP systems, eProcurement systems, and supplier ecommerce applications. For example, a manager creates a requisition order and emails it to her superior, who approves it and forwards it to a procurement professional, who enters it into a supplier’s ordering system. Transcription errors are among the most common, so perhaps the order is entered as 10 units instead of 100.
Consequently, the buyer receives a tenth of the products required to complete a production run, damaging their relationship with a key client and costing them business.
In the Harvard Business Review, Thomas C. Redman wrote about a project AT&T instigated to reduce invoicing errors. The project found that 40 percent of invoicing data contained errors leading to the company overpaying by tens of millions of dollars. The majority of these errors were caused by incorrect data entry somewhere along the chain from the supplier to AT&T’s accounts payable department.
Other costs associated with data entry errors are less tangible but no less significant. A business that can’t trust the data in its eProcurement or ERP platform can’t use it to make sound decisions. If they do use it as the basis for decisions, those decisions are unlikely to produce the best outcomes. Bad data is a poison that impacts competitiveness and productivity.
A business that chooses not to live with faulty data is forced to fix it, the 10 in 1–10–100. Remediation of data errors is expensive for a couple of reasons. It’s much harder to automate remediation than prevention, and remediation is often a massive project dealing with years of historical data.
In the AT&T project we referred to earlier, the only way to discover how much invoice data was incorrect was to go back to the original documents and painstakingly compare them to the data that was entered into the payment system. For a company of any size, that’s a huge amount of manual work, which has to be double-checked because it’s prone to the same errors that caused the problem in the first place.
Finally, we come to the 1 in 1–10–100. The least expensive way to deal with the errors caused by manual data entry is to prevent them from ever occurring. By automating the flow of data between buyers and suppliers, the likelihood of incorrect data finding its way into eProcurement systems is minimized, if not eliminated.
Historically, the integration of eProcurement and ecommerce applications was an expensive and time-consuming process. Organizations only integrated the biggest suppliers for features such as punchout catalogs, requisition order, purchase order, and invoice automation. That left a long tail of smaller suppliers that represented a lower dollar value but often a much higher number of total invoices.
Modern cloud integration platforms facilitate quick and low-cost integration projects, enabling buyers to integrate a far larger proportion of their supplier-base, automating the flow of data between buyer and supplier and drastically reducing data entry errors.
Prevention of data entry errors through automation has always been less expensive than remediation or ignoring the problem, but today’s integration technologies make automation the only reasonable choice.
About the Author: Brady Behrman is the CEO and founding partner of PunchOut2Go. As an entrepreneur with experience and proven track record in building technology businesses that focus on client success innovation, Brady and his team help organizations of all sizes around the globe adapt to the ever-evolving, complex B2B Commerce & eProcurement technologies.