AppZen provides various models to help audit expense reports so auditors can focus on the high-risk expenses.
1. Amount Verification
Ensures that the amount submitted for an expense line is below a specific threshold, broken down by cash (personal card or cash currency) and credit (corporate card). This is usually set to a high number like $10,000 (often only to be violated by marketing spend).
2. Amount Verification by Expense Type
Checks specific expense types and ensures that the submitted amount is below a specific threshold, broken down by cash (personal card or cash currency) and credit (corporate card). If an expense line is flagged as low risk but the user amount is over the threshold it will be flagged as high risk. Amount verification by expense type checks if there are attendees associated with the expense. The rule will always divide the total amount by the number of attendees and will use that against the threshold.
3. Anti-Corruption and Bribery Check
Performs an FCPA check, where AppZen looks at different databases (e.g., Government, FBI most wanted, etc.) to match attendee names and business titles that are on those lists with names submitted within an expensed meal. Lists are US-based.
4. Receipt Verification
Checks if the user input data matches the receipt data picked up by AppZen's OCR. Validates whether:
- There is a receipt at all
- The amounts match between the user input data and the receipt
- The tip amount falls outside a specified variation set in AppZen (e.g., greater than 25%)
- Matches the date on the receipt against the user input date. The date has a two-day variance from the actual date. For example, if the date on the receipt is Jan 1st 2019 and user input date is Jan 3rd 2019 AppZen will not flag as high. However, if the receipt date is Jan 1st 2019 and user input is Jan 4th 2019 AppZen will flag as medium according to the receipt verification rule.
Please note: Certain Appzen expense types do not require a receipt: Tolls, Mileage, and Gratuity
5. Credit Card Check on Cash Lines
Checks whether the employee used a personal credit card to pay for an expense when a corporate card was expected.
6. Daily Meal Limit
Checks meal amount across expenses & expense reports, against the amount a person can spend on meals in a specific day. Specific meal expense types (e.g., breakfast, lunch, dinner, snack) can also be broken down further to set specific limits per type.
7. Duplicates within a Report
Analyzes receipts for the date, amount and merchants to find duplicate claims within a report. A side by side image of the duplicate receipt will be visible to the auditor for review.
8. Duplicates Across Reports
Duplicates Across Reports searches across the reports of the employee who submitted the expense, as well as across all other employees' reports to detect any duplicates checking date, amount and merchants. The Duplicate Check beta OCR logic is built into Duplicate Across Reports in regards to searching an expense with the same receipt image. A side by side image of the duplicate receipt will be visible to the auditor for review.
9. Expense Age Check
Looks at the date on user input & the receipt to validate if the submitted report is older than a predetermined duration (e.g., 90 days or 180 days).
10. Merchant Category Check
The merchant is searched on the web to verify if the merchant is associated with any questionable activity including adult entertainment, movie theaters, spas, golf, etc. MCC Codes are used (AppZen also looks for MCC codes in Credit Card transactions)
Additional logic has been implemented with this model that will improve on nuances in variations of the merchant name. Such as the word, "store" is the merchant category name "stores" (with an s) will not be flagged. However with recent improvements "stores" will be flagged as high risk.
11. Receipt Itemization Verification Check
12. Restaurant Price Check (Merchant Price Check)
Perform an internet search (e.g., Yelp) to verify the average amount that a meal should cost for an individual, and validate it against the cost and number of people an employee has submitted within their expense report.
13. Unauthorized Expenses
This is a keyword search based off of different unauthorized categories in a list of lexicons (e.g., Spa, Alcohol etc.). With recent updates to AppZen, functional admins can configure the model for "Business Class" and "First Class" expense types.
Now the functional admin can separate the two unauthorized categories out to possibly add a different threshold between the two unauthorized categories.
14. Weekend Expenses Check
Looks at the date on user input & the receipt to validate if a purchase was made on a Saturday or Sunday (or another designated day).
15. Merchant Black List
This model identifies spend at specific Merchants. A blacklist is created with a list of merchants that an employee is prohibited from using. This model is based on user input details. If the user doesn't input any merchant, Appzen will use OCR to find the merchant on the receipt. If an expense from that list of merchants is found, the line will be flagged as high-risk. With improved logic with the Merchant Blacklist model, we have made the model more accurate when checking for merchants on a receipt.
If the merchant name was on the list to be blacklisted for example, "Burger King" it would flag as high risk. However, if the merchant name was "Burger King" the model will not flag as high risk. This is due to the spaces in the keywords. Now with the improved logic, "Burger King" (with spaces) will be flagged as well.
16. Healthcare Professional Check (HCP)
Checks if medical professionals are listed among business attendees. AppZen utilizes its Healthcare Professional Check (HCP) to cross-reference the US National Provider Identifier Registry (NPI) to spot expense reports that have a healthcare professional tagged as the attendee on an expense report. If the medical professional is marked as ‘HCP’ in the expense management software, this model is not triggered.
Here is a workflow on how HCP works:
17. Non-Conforming Receipt
The Non-Conforming model will determine if an employee uploaded a valid receipt. Some examples of non-conforming receipts are handwritten receipts, sticky notes, random photos, blurry receipts and receipts with white space. If the receipt makes up less than a quarter of the full page AppZen will flag this as high. Depending on the confidence level the receipt can be marked as high or medium risk.
Exceptions for Non-Conforming Receipt Model:
Receipt Verification - The Non-Conforming receipt model is tied to the receipt verification threshold model. If the receipt is below the threshold and the receipt qualifies as a non-conform receipt it will not flag as a high or medium risk. However, if the receipt is above the threshold limit and it qualifies as a non-conform receipt, the non-conforming receipt model will flag as either medium or high-risk.
Parking Tickets - Parking tickets are not subject to the Non-Conforming receipt model at this time.
Examples of Non-Conforming Receipts: