form on a top of a colored shape background.

Guides

How to automate annuity statement processing using OCR and Impira AutoML

Downloading PDF annuity statements from various carriers and retyping the data into Excel is a tedious and time-consuming process. This guide describes how financial advisors can set up Impira AutoML to convert all of their fund documents into a spreadsheet instantaneously.

Download sample files

Using automated machine learning (AutoML) for annuity statement processing

Financial advisors, RIA firms, CFP practitioners, or anyone who provides investment guidance is familiar with the multitude of documents that must be managed on behalf of their clients. Carriers typically provide a web portal where statements can be downloaded as PDF files. From there, the financial information (e.g., beneficiary name, gain/loss, beginning and ending values, etc.) must be manually re-keyed into a spreadsheet so that the information can be used for forecasting or sent on to a CRM system.

In the past, brokers have attempted to use optical character recognition (OCR) technology to expedite the process, but found that the results are less than satisfactory. Today, however, advancements in machine learning technology make it possible to cleanly extract the data.

Impira AutoML builds on the vision of OCR technology by applying an additional layer of intelligence to the data capture process. The system is designed to reduce the number of errors that traditional OCR generates by allowing users to teach the AutoML to learn from its mistakes, quickly and continuously improving. Our easy-to-use interface mimics the look and feel of a spreadsheet, which means that anyone can use the tool and immediately realize productivity gains.

With Impira AutoML, you can automate your financial statement processing in less than five minutes, and scale the number of clients that your firm can manage.

The business benefit

More accurate than OCR, Impira AutoML makes it easy to convert annuity statements into a spreadsheet format. This means that financial advisors can dedicate more time to advising, and spend less time manually copying and pasting information from one document into another. The ability to work with a wide range of carrier documents and the ability to adapt to changes in those documents over time will ensure that brokers are working at their full potential. Delight your clients with fast turnaround times, and increase the number of clients your firm can manage. Get started with Impira AutoML today and experience how easy it is to automate your document-heavy processes.

Step 1: Sign up for a free Impira account

If you haven’t already done so, head on over to impira.com/signup to create your free account. 


Step 2: Create a new collection

Once you’ve created your account, go ahead and create your first collection by clicking the + symbol on the left-hand sidebar.


In the prompt, give your collection a name, like “Stark Statements.”


Step 3: Upload your statements

Let’s upload some sample statements into your account:

  1. First, make sure that your collection is selected.


  1. Drag and drop your files onto the page.


Tip: There are several other ways to upload files into Impira. You can also:


Step 4: Train an AutoML model on your statements

Notice that your uploaded files have arranged themselves into a table, and each row is a different file. So far, our table only has one column (“File”). When you train AutoML, the fields that you extract will be displayed as additional columns on this table. Let’s open up the first file to begin training the AutoML.

Extracting initial data from your forms
  1. Double-click on the first file to get started.



  1. Let’s start by extracting the Investor Number. Click and drag to highlight the number. On the right-hand sidebar, give the field a name (e.g., “Investor Number”) and click the Add field.


  1. Close the window by clicking the X in the very top-right corner.

  1. This’ll return you to the collection view, where you’ll see a new column in the table for “Investor Number.”  In the column header, the spinning indicator informs you that the AutoML is actively learning as you label items. It’ll then immediately make predictions for the Investor Number across every document in the collection.



Step 5: Fine-tuning Impira

Some cells may initially be blank, or have a colored flag displayed on the left side of the cell. This flag color is a visual confidence indicator of Impira AutoML’s prediction. A green marker indicates high confidence, and a red flag suggests that you review this prediction.

Because Impira AutoML is a continuously learning system, you can boost these confidence levels by confirming correct information or correcting misidentified information. These affirmations increase Impira AutoML’s accuracy by teaching it what you want it to do. The more it learns, the more you’ll see red flags turning green. Read more about how reviewing your predictions custom-tailors Impira to your needs.



Double-click a red-flagged cell and take a look to see if Impira AutoML identified the right information you want. Click the check mark on the right-hand side if it’s correct.

Sometimes, Impira AutoML will misidentify a piece of information (e.g., accidentally highlight the wrong address). Easily correct this by clicking the X in the right-hand sidebar and use your mouse to redraw a box around the correct address.

Repeat with more fields

Repeat Step 4 to create additional columns for all of the fields you’d like to extract (e.g., End Date, Beginning Balance, etc.).



Now that your collection is configured and performing well, you can upload additional statements to the collection and see them added on as new rows. The AutoML fields that you’ve already defined will automatically apply to the new files.

Export your data

When you are satisfied with the results, you can easily export the data from Impira. To download the data as a CSV file, click the “Download” button in the upper right-hand corner. In the drop-down, select “All files records (CSV).”
When you’re satisfied with the results of your data extraction, you you have a few options for exporting your data:

Option 1: Export via CSV

Click Download in the top-right corner and select As CSV.


Option 2: Access your data via API or IQL

You can also access your data through an API, and you have the option of utilizing Impira Query Language (IQL) to query specific data, giving you as much control and flexibility you need over your data. This document will also show you how to utilize a Poll API.

And if you're feeling a bit more adventurous, you can always connect a webhook with Impira.

Get started in minutes.

Already using Impira? Sign in.