How Much Does an Invoice Data Extraction Cost?

On an everyday basis, business executives have to keep up with the changes happening in the business dynamics of how businesses operate. In accomplishing so, they spend quite a lot of their time contemplating the adoption of the latest technologies.

No matter how much the executives vouch for their gut feel and instinct, the reality is that very few executives can have their way without a clear objective analysis of the investment needs that justifies the budget spend.

What’s a TCO?

Total Cost of Ownership (TCO) and Return on investment (ROI) are two of the widely used approaches in securing a budget for planned initiatives.

Total Cost of Ownership is a calculation method that determines the overall cost of a product or service throughout its life cycle. TCO is not just a cost figure on paper. It should inform relevant stakeholders of the following:

1. Uncertainty and Risk factors concerning the initial upfront investment or setup costs.
2. Strategic fit and Competitive advantage of the selected set of technologies.
3. Cost overruns that could materialize due to implementation-related risk factors such as delay, tools, and techniques, skills availability, etc.
4. Comparative assessment with the set of technologies that are being replaced.
5. Process cost on a per-unit basis (per invoice basis in the case of Accounts Payable).
6. Workforce expense on a basis (per invoice basis in the case of Accounts Payable)

Cost Drivers

The types of Cost we‘ll consider are:

• Direct costs are nothing but the workforce-related expense in terms of the FTE Load and are derived from keystroke count. The Full-Time Equivalent (FTE) Load is the effort equivalent to one employee working full time on data entry tasks. For example, if the number of employees is 9, then the working hours for each employee will be 9 hours, and if the efforts in hours of each data-entry operator are equivalent to 3 hours, then the data entry FTE Load is 3.

Here are some key measures to look at:

1. Average Number of Keystrokes per invoice
2. Average Typing Speed (Keystrokes per unit time) — (Continuous typing is not possible. A multiplier that acts as a buffer for all supplementary activities performed by the operator and is incidental to the invoice extraction process should be used e.g. time for document movement, locating fields, comprehending difficult to read data, etc. The multiplier can be determined empirically or by basic statistical analysis.)
3. Average wages for Data-Entry Work per unit time based on FTE load.
4. Average overheads for Data-Entry Work per unit time.
5. Indirect costs (These are additional effort associated with solving impediments, issues, and incidents such as data entry errors, duplicate payments, etc.

Some key measures are:

1. Software downtime and troubleshooting expenses.
2. Average wages for Data-Entry Work per unit time based on FTE load for rework.
3. Average overheads for Data-Entry Work per unit time for rework.
4. Hidden (intangible) costs (include penalties for late payments, vendor issue escalations, employee rotation, etc. This component should be based on the firm’s historical data.)

The above describes the costs involved in a typical manual invoice data extraction process. There are additional costs for an OCR-based invoice data extraction process.

Data extraction systems based on OCR engines generally make use of templates for invoice-data extraction. A template is either a zonal OCR with fixed page locations for individual data fields or a rule-based OCR engine extractor configured with a sequence of if-then rules telling the software where to look for specific information and a parser to make sense of that information.

There are additional cost structures that come into play for a data-extraction system set up in addition to direct and indirect costs due to employee wages

1. Cost of the OCR system: Costs will vary from vendor-to-vendor, implementation-to-implementation to organization-to-organization.
2. Licensing Cost: Licensing structures vary from vendor to vendor. Almost all structures follow a yearly renewal process. It can be a fixed upfront cost or a pay per use ($/page, $/field, $per document type) arrangement, or a combination of both. It can increase with higher functionality. In general, an AI-powered cloud-based solution may be more costly than an on-premise template-based OCR data-extraction system.
3. One-time set up cost: Initial set up is time-consuming, requires management involvement, appraisal of the extraction system with vendors, IT-related enablement (if on-premise), and is thus costly. Integration, system configuration, and setup, template configuration is also quite necessary.
4. Cost of resets as and when is required to account for changes in layout and structure: Creating and deploying configuration files, customizing extractor, configuring parsers, and pre-validating using training sample sets are a few activities to which costs can be attributed to depending on the type of data extraction system chosen.

• Training costs
• Maintenance

Note: Even with automated data extraction systems, for some organizations, a good percentage of invoices will still have to be processed manually.

Insufficient volumes of a vendor invoice might not justify the creation of a template for automated data extraction. As in manual data extraction, Indirect Costs due to verification and rework also must be accounted for. The hidden costs are usually the costs for managing the external consultants that maintain the OCR software.

The more the errors, the more the cost of data extraction. This is exactly where TCO estimates go wrong often. Delays and inaccuracies also lead to cost leakages. Organizations must do a great deal to ensure enough pre-processing for a smooth data extraction process with minimal errors. There may have to be discussions on invoice layouts, field descriptors, page background, font size, no of pages, etc. with the intent to improve the accuracy and efficiency of the automated data extraction processes. It is, however, difficult to measure and factor in such costs.

Key KPIs that must be monitored for the Accounts Payable Process

1. Invoice exception rate: Data inaccuracies, omissions, and inconsistencies are just a few of the many causes of invoice exceptions, which affect the Accounts Payable processes.
2. Average invoice processing time: Time is money. The longer it takes to process invoices, the more it will cost. Wastages should be avoided, and exceptions managed.
3. Days sales outstanding (DSO): The number of days it takes to pay a supplier after receiving their invoice is another important metric. Vendor relations and subsequent payment discounts are at stake here.

While some organizations may skip the TCO calculation altogether citing it as a simple heuristics exercise, TCO information is more vital than one may agree. It is very crucial to the successful implementation and monitoring of a data extraction system that can be used to automate downstream Accounts Payable Processes. We leave you with a sample TCO template.

Find out more on Neutron that has features like customizable extraction engines and flexible templating to make it a best-in-class invoice data extraction system.