Deduction Finder

Now that you've connected your Plaid Integration with Hurdlr's Expense Tracking API, you're ready to start leveraging all of the Expense Tracking features that Hurdlr has to offer, including Deduction Finder.

1. How it works

Through Hurdlr's Plaid Integration, your user's transactions are pulled into Hurdlr as soon as Plaid makes them available. All of those transactions are then run through Hurdlr's proprietary algorithms, which leverage the user's business type, accountant best practices, and the behavior of similar businesses within the Hurdlr app's 700k+ users. The expense transactions are then categorized in three primary ways:

a) Business Categories: Hurdlr uses categories specific to the user's business, which a non-accountant business owner can understand. For example, an Uber driver's expense might be categorized as "Car Wash".

b) Tax Categories: All expenses are also mapped to a tax category, which ultimately maps to specific line items on the 1040 Schedule C, 1065, and 1120S.

c) Deduction Confidence: Based on the user's business type and many other factors, Hurdlr's algorithms predict whether any given expense is deductible.

2. Utilizing business type to provide accurate predictions

Of course, what is deductible for an Uber driver is very different from what is deductible for, say, a freelance software developer, and for that reason, Hurdlr's algorithms weigh the user's business type very heavily.

With transactional data and behavior from 700k+ users of its apps, all segregated by business type, Hurdlr's algorithms are built on top of a more advanced dataset than any other deduction finding algorithm available. As the economy evolves, and user behavior changes, so too do Hurdlr's algorithms.

When you are registering your user, be sure to include an accurate bizType attribute, as that will unlock the true power of Hurdlr's algorithms.

3. Leveraging all available transactional data

Since Hurdlr's algorithms are built on top of your Plaid integration, all the data that Plaid makes available (merchant name, transaction description, category, etc.) are all automatically leveraged in Hurdlr's deduction confidence algorithms. Neither your developers nor your users need to have any tax expertise. Your team will not need to build any transaction or category mappings; you get all of that automatically through Hurdlr's existing Plaid integration.

And, if you use a custom data source outside of Plaid, the Hurdlr API team would be glad to help you securely expose that data in a way that the Hurdlr API can consume it. The Hurdlr API team has established custom interfaces with banks and other financial aggregators, making the integration process turn-key for our partners. Don't hesitate to email us at [email protected]

4. Getting the user's expenses

First, you should retrieve a list of the user's expenses, which are expense transactions that your user has not yet approved.

5. Identifying a transaction's deduction confidence

On each expense transaction, you will see a business-specific expense categoryId, as well as a nested taxCategory object.

You will also see a confidence field, which can have one of the following values:

Value

Use Case

LIKELY

Expense transaction is likely to be deductible for the given user; you may want to display all LIKELY transactions to your users and allow them to approve them in bulk

QUESTIONABLE

Expense transaction needs user review prior to being classified as business or personal.

UNLIKELY

Expense transaction is likely to be personal (non-deductible) for the given user; you may want to display all UNLIKELY transactions to your users and allow them to approve them in bulk

6. Taking advantage of Hurdlr's learning abilities

Whether you are relying on Hurdlr's deduction confidence or allowing your users to approve/reject Hurdlr's recommended expense classifications, Hurdlr's algorithms get smarter over time, automating more and more of your users' actions.

Hurdlr's A.I. suggests expense rules for each specific user, which can be easily fetched:

curl \
  --request GET \
  --url https://sandbox.hurdlr.com/rest/v5/expenses/rules \
  --header 'Authorization: Bearer ${access_token}' \
  --header 'Content-Type: application/json' \

The response from GET /rules contains an array of the user's expense rules:

{
  "data": [
    {
      "id": 1118497,
      "transactionName": "Starbucks",
      "matchType": "NON_EXACT_NAME",
      "autoClassify": "BUSINESS",
      "suggestedStatus": "PENDING",
      "categoryId": "13325613",
      "personalCategoryId": null,
      "vendorId": 237792,
      "businessId": 416080,
      "lastUpdatedDate": "2021-08-31T22:39:18.000Z"
    }
  ],
  "lastUpdatedDate": "2021-10-07T17:45:42.424Z"
}

On each rule, you may find the following attributes to be of particular interest:

Field

Description

Format

id

Id of the rule record

Numeric

transactionName

Name of the transaction to determine whether it matches the rule

Any string

matchType

Type of matching algorithm applied

Must be one of the following: "EXACT_NAME", "NON_EXACT_NAME"

autoClassify

Type of auto-classification

Must be one of the following: "OFF", "BUSINESS", "NOT_BUSINESS"

suggestedStatus

Status of the A.I.-based rule suggestion

Must be one of the following: "PENDING", "ACCEPTED", "DECLINED", "DELETED"

In order to accept a rule that Hurdlr's algorithms have suggested, simply update the suggestedStatus to "ACCEPTED":

curl \
  --request POST \
  --url https://sandbox.hurdlr.com/rest/v5/expenses/rule \
  --header 'Authorization: Bearer ${access_token}' \
  --header 'Content-Type: application/json' \
  --data '{
    "rule": {
      "id": 1118497,
      "suggestedStatus": "ACCEPTED",
    },
  }'

Now, all of that user's new expense transactions that meet the suggested rule criteria will be automatically classified and accounted for in the user's tax estimates and tax reporting.

7. Embeddable User Interface

If you want to get up and running quickly with a proven Deduction Finder UX that matches your colors, fonts, and branding, then the Embeddable User Interface may be your best bet.