Automatic Digitization Of An Invoice
Accounts payable (AP) process of a company plays a critical role by managing the payables i.e invoices owed to their customers. The process involves timely receiving the invoices, verify the accuracy of the data, validating the data and processing the invoices efficiently and effectively.
As a saying “Time is essence to any business” we believe Accounts Payable stands top in performing the tasks on time as late payments incur additional charges or early payments discounts.
Traditionally, invoices are obtained either on paper or electronically in the form of e-invoices. Invoices are processed manually. An AP department employee validates the credentials of the purchase and the seller in the invoice, checks for possible duplicates, enters the fields in each invoice into the company’s Enterprise Resource Planning (ERP) system before processing the payment.
Depending on the scale of the business, a company may receive from a few hundred invoices per month for a medium-scale company to several thousands of invoices per month for a large-scale company. Assuming that on average an employee can process 5 invoices per hour i.e. about 40 per day and between 800 – 900 invoices per month, a company will have to maintain several employees for this purpose depending on the scale.
Moreover, this whole sequence of tasks is tedious, prone to human error that may snowball if not detected early and invites unwarranted loss to the company. An automatic invoice digitization software can help reduce the manual effort, time and cost involved in this process.
The process involves uploading an invoice into the software with the given mandatory fields, values and tables needs to be extracted. The fields that are extracted are validated against the respective data which will be used for validating and processing the payment either by a human or an ERP software.
The Current State Of Accounts Payable And Its Implications.
Making sure payments are made in time is one of the most important tasks of the AP department. Late payments negatively impact a company’s credibility, reputation and attract monetary loss in the form of late payment charges. This 2017 Simens’ report states that UK small and mid-size enterprises (SMEs) miss out on over £250 billion of liquid cash flow because of slow and late payments. This 2017 survey report by Plum Consulting based on 3000 companies across 11 countries states that 11% of all invoices issued by SMEs are paid late and impact SMEs adversely. In the top 3 reasons given for late payments by paying companies, “No Reason” was the answer 20 –40% of the time. This is suggestive of operational inefficiency. The Atradius payment practices barometer report of India, 2022 states that 54% of the total value of Business-to-Business (B2B) invoices are paid late. The main reason for payment default by businesses in electronics/ICT and chemicals industries was due to administrative delays in the customer’s payment process. These reports suggest that the time taken to process an invoice may play a key role in late payments. The Atradius payments practices report of the US bolsters this argument. It states that, “Late payments were most often due to administrative inefficiencies in the customer payment process, according to 48% of companies polled in the US.”.
Covid-19 had a significant impact on the state of AP. This 2021 report by Ardent Partners that captures the perspectives of 187 AP, finance and Procure-2-Pay (P2P) professionals finds that 94% of all AP departments were impacted by Covid-19, with 9% of the leaders defining the impact as “massive” and 22% defining it as “high”. The impacted organizations sought higher levels of automation and 2/3 of them now believe that their AP team is either “very valuable” or “exceptionally valuable” and see the need to invest in the improvement of AP. Here are their top priorities for improvement:
- Improve AP reporting and data analytics – 55%
- Eliminate paper invoicing and reduce manual tasks – 41%
- Enable more suppliers to submit e-invoices – 36%
- Eliminate paper check payments – 30%
- Implement AP automation – 29%
- Reduce processing costs – 27%
And here are the top challenges AP faces:
- Invoice/payment approvals taking too long – 57%
- High percentage of exceptions – 45%
- Too much paper – 28%
- Manual B2B payments – 22%
- Late supplier payments – 20%
These reports suggest the increasing need to invest in the automation of AP. Automatic invoice digitization is one of the key steps in AP automation as this is an area where a human employee would a lot of time and effort. Where a human might take several hours to perform a task, a software can do it in minutes and humans can be employed merely to handle the exceptions thereby reducing time, effort and cost.
Process Flow Of Automatic Invoice Digitization.
The key features such as data extraction, data validation make us outstand in the current market. Here is the typical process flow for automatic invoice digitization:
- Uploading of invoices either single file or in bulk through the web interface or the API interface offered by the tool. Image formats such as .jpg, .png, etc. and pdf formats are generally supported.
- Extraction of text and tables from the invoice(s). Text extraction is done using optical character recognition (OCR). Table extraction is done by localizing the tables, identifying the rows and columns and extracting the text from the cells using OCR. OCR and table extraction typically use deep learning models. Deep learning models report an accuracy measure that indicates how well the models can perform the desired task.
- Identification of the various fields in the invoice. Once the text and the tables are extracted, the next task is to identify what text is what. For example, in an invoice, one may need to extract several fields such as the invoice number, the invoice date and total. Every tool extracts the common fields found in an invoice and typically have provisions to extract custom fields. Different tools in the market use their own algorithms to do this and report a performance measure.
- Manual handling of exception cases and verification. A human verifies the information extracted from the invoices and manually enter the data of exceptional invoices that the tool has not been able to extract.
- Digitized invoice information is sent for further processing into the organization’s ERP system and analytics.
This process saves time as the software tools can process a single invoice in anywhere from a few seconds to a minute. Manual effort in data entry is reduced as a human is required only to handle exception cases and verification. These software tools generally have integrations with an existing ERP systems that an organization may already be using, for seamless integration into the AP process flow. Furthermore, the software tools might offer extra features such as support for other forms of documents e.g. receipts, purchase orders, legal identification documents, etc; support for non-English languages, etc.
Why iKapture
iKapture focuses on versatility, user-friendly solution that can customize based on the customer needs. It offers the following features that makes us outstanding in the market.
- Easy addition of custom fields.
- Allows to customize fields based on the specific needs of the business. For example, the fields and identification numbers for businesses vary from country to country and the software needs to capture this. Typically in the market the softwares needs additional efforts to customize which will be time consuming where as IKapture offers a simple in-built solution to do this. One can add custom fields into the software at any time by simply naming the field and providing atleast one synonym to it through the simple interface provided for custom field addition. The custom field will be extracted from thereon.
- Easy to retrain for the extraction of new layouts of tables. Tables in invoices can be quite complex and can have several different kinds of layouts. For other tools, if a table with a new kind of layout the data extraction is limited, one may need to precisely crop the whole layout of the table manually for a few documents so that the deep learning model of the tool is “retrained” to identify the new layout. IKapture on the other hand needs the header and the footer of the new table layout to be captured manually instead of the whole layout and the whole table layout gets automatically extracted.
- Easily customizable to new kinds of documents, iKapture offers versatility in this scenario as it can extract from any kind of document as long as the document has a key-value format for the fields. This makes customization to new documents virtually the same as addition of custom fields as described above.