OCR (Optical Character Recognition) has slowly become the norm to make receipt scanning easier for expense management. Now employees just have to click a picture of the receipt, and the scanner extracts most of the data.

But automated data extraction comes with its caveats. It's important to consider how the process impacts your data's security, and how an inefficient extraction system can hinder your entire expense management cycle.

How data extraction ties to privacy

As Finance, when you’re implementing an external platform, you’re concerned about data safety. This depends on the manner in which data extraction is implemented by the company. OCR is used to read data from images, and convert it into text. But not all the text scanned from an image is relevant; there needs to be a mechanism to filter out the important data and enter it in the expense form. Some companies, like Fyle, implement automation: OCR, coupled with Machine Learning to extract relevant data. Some others outsource the job to people - this means random third-party workers are privy to your data.

‍

The impact of inefficient data extraction

• If data extraction is incorrect, what’s the point? Many expense software have built-in OCRs, but the data extracted isn’t accurate. The employee ends up correcting all the mistakes in the expense form manually, which still takes time. A complicated or error-prone process discourages employees from filing their expenses on time.
• It’s slow - From several minutes up to an hour, some software don’t really speed up data extraction (especially if it’s being done by people, it’s bound to be slow). You scan a receipt and are told that you’ll be notified when the extraction is complete; which means your employees are not exactly filing expenses as they go. If you interrupt the process to add some information yourself, the extraction is stopped.
• It identifies just the basic information - Most receipt scanners are limiting, extracting only basic line items like currency, amount and merchant.

Fyle’s upgraded process for receipt scanning

Fyle’s ML-based data extraction: Your data is yours

Fyle uses ML to extract relevant data from a receipt and auto-populate the fields. The process isn't just fully automated, but also ensures data privacy. None of your data is outsourced or available for any external person to view.

Speed + Automation

Your employees will truly be able to expense as they go, as Fyle’s data extraction system will enable them to scan, create, and report expenses quickly. Since it’s fully automated through ML, the system extracts data within 2-3 seconds of scanning your receipt. You can then review and submit your expense immediately, without spending time filling out expense fields. Additionally, your employee expense reimbursement cycle becomes much faster.

Accuracy comes first

Dealing with a not-so-smart OCR can be frustrating, where you have to manually cross-check and edit data that has been incorrectly extracted. To resolve this, Fyle identifies and maps multiple fields correctly. The more you use it, the more the machine learns. You don’t have to worry about mismatches and inefficiency.

OCR that auto-populates most of your expense form

You can see data auto-filled for fields like currency, amount, date of spend, location, category and merchant. Just state the purpose of the expense, review the information, and you’re good to go.

Automatic credit card transaction matching

Matching credit card transactions with expenses is tedious, time-consuming and laborious, but necessary. Fyle’s ‘auto-match’ feature ensures that your card transaction is automatically matched with an expense created, as soon as the data extraction is complete. So it's accurate, fast and doesn’t compromise on employee happiness.

The data extraction is seamless no matter where employees file receipts from

• Fyle’s mobile app - We know that handling paper receipts is chaotic, especially for the employee. Fyle’s automated receipt scanner, ‘Instafyle’ helps address this problem, while remaining incredibly easy to use. A user only needs to click a picture of the receipt via the mobile app; and the system will extract the data and auto-populate most of the expense form fields.

• Bulk uploads - If you have employees that prefer to collect all their receipts and file them in bulk, they can still reap the benefits of automated data extraction. Bulk filing can be done one of two ways - rapidly click pictures of multiple receipts; or drag and drop saved bills on the webapp. Either way, Fyle’s OCR will extract and fill in the information in mere seconds.

• Gmail & Outlook 360 - For e-receipts, users can record expenses without leaving their inbox. This means they don’t have to juggle multiple applications at once. The Fyle Gmail & Outlook 360 plugins scan and create expenses with a single click. As soon as you choose to ‘Fyle it’, the data is pulled from the receipt, and the attachment is imported into Fyle.

• Email forwards - If users find attaching receipts themselves too tedious, they can also forward their receipts to receipts@fyle.in (or set up auto-forwarding!) and Fyle will automatically create the pre-filled expenses. This is especially useful when you receive receipts from specific emails or vendors, for e.g. Uber receipts that come from payments@uber.com.  

• On the webapp - For your employees that prefer the webapp, they can drag & drop receipts on the computer, and Fyle’s ML works its magic.

‍

Automated data extraction doesn’t mean you lose control

The automation exists to make life easier, but you can easily review and edit expenses before submitting them. To add to it, the expense form can be customized according to your needs. So instead of scrambling for information from employees, you can capture additional data, or make certain fields mandatory with ease. Fyle’s workflow adapts to your expense management process, and not the other way round.

Duplicate Detection

For most organizations, duplicate expenses are a big compliance problem. It could be a result of a mistake or an indication of possible fraud, which can be serious. With ML, Fyle can detect duplicates in real time, when employees scan their receipts. If fields like date of spend and amount are found to be repetitive, the employee is notified of a possible duplicate.

ML, coupled with real-time policy checks

If you have expense rules or policies in place, Fyle will automatically run these policy checks in real-time, as and when employees are filing their expenses. If there are violations, the expenses will be flagged. This reduces the risk of fraudulent activities, duplicate expenses and overspending. This also means auditing happens at the first level, so you can avoid errors and compliance violations in your workflow. Even if you have complex rules, Fyle’s advanced policy engine can handle it!

‍

Additional automation: auto-fills

If you want your expense fields to resort to a certain default value based on your recent selections, you can enable ‘auto-fills’. Every time you create a new expense, the line item is pre-filled, saving you effort and time. For instance, if you’re working on a particular project, the last selected project will be auto-selected for all your expenses.  

Unlimited receipt storage

Traditionally, receipt maintenance is a document-intensive process, where you manually extract and store the data in a database. If you wanted to search for receipts, you’d have to put in some effort. As you adopt intelligent data extraction with Fyle, the aim is to give you a well-rounded automated receipt system. You have access to an unlimited receipt storage (with a powerful search functionality) - so you can filter and find data on demand.  

The future of expense management

At Fyle, we understand how tedious expense management is, not just for finance teams, but for every stakeholder involved. That’s why we’ve built a platform that aims to help you spend as little time and effort as possible on it. With an automation-first approach and focus on compliance, we fit into your expense workflow. We’d love to show you around, schedule a demo with us today