Invoice capture is a growing area of AI where most companies are making their first purchase of an AI product. This is because invoice capture is an easy to integrate solution with significant benefits.
While digitization helped automate numerous processes, mostly rule based software was used in digitization. Invoice capture software is different. Invoice capture involves both reading the invoice text with Optical Character Recognition (OCR) and understanding its context with machine learning.
We answered all your invoice capture related questions:
What is invoice capture?
Invoice capture (also called invoice data extraction or invoice OCR) is extracting structured data from invoices so invoices can be automatically processed. Invoice capture has been the first back office process to be automated with AI for most companies.
This is only relevant for invoices that are received outside of an Electronic Data Interchange (EDI). Invoices that arrive via EDI can be auto-captured since they are already in the form of structured XML files. For more on different types of invoices, feel free to read our article on invoices.
If there is significant uncertainty about the data, a human is notified to take a look at the invoice. If data extraction is deemed to be successful, data is fed to the record keeping and payment systems.
Companies need to set up quality assurance processes in any automated process where errors can be costly. Invoice capture is no exception. To ensure that wrong payments are not made, suspicious invoices and invoices that require payments beyond a certain limit would need to be reviewed by humans.
What are the benefits of invoice capture?
- helps reduce back-office costs by reducing manual effort
- allows employees to focus on higher value added activities
- reduces invoice processing errors
- allows faster turn-around time which prevents the unnecessary back and forth between suppliers and the business which consumes valuable employee time
- allows auditability: Invoice data can be stored with bounding visual boxes that show where data was extracted from the invoice. In case the company discovers that data extraction had faults, these documents can be used to understand the source of errors which can be corrected for future invoices.
- improves compliance: Invoices hold numerous data fields which are traditionally not captured manually. By capturing all data on an invoice, invoice capture software enables companies to run compliance checks on invoice data.
What are the differences between invoice capture and OCR?
While OCR captures text, invoice capture solutions capture key-value pairs and tables which are required to auto process invoices.
Capturing key value pairs
Invoices include key value pairs such as company name, bank account number etc. Invoice capture solutions can extract key value pairs from documents.
Most invoices include an itemized list of services or products provided. Invoice capture solutions can recognize these itemized lists and process them.
What are different types of invoice capture solutions?
In the invoice automation landscape, there are 3 types of solutions:
- Template based solutions: End-user inputs the document structure to the software. These solutions were prevalent before the recent rise of machine learning solutions. However, they are no longer relevant since
- There are many different structures for invoices and these structures tend to change over time. This results in errors.
- Using templates creates a code base that needs to be maintained
- Inputting template structures to the software is additional work. Ideally, automation solutions should not create new manual tasks for users.
- Pre-trained machine learning (ML) solutions: Companies build automation solutions based on millions of invoices. These solutions are great however can run into issues when they face types of invoices they have not encountered before.
- Continuously trained ML solutions: Best solutions on the market. They are trained on millions of invoices and developers work with their customers to ensure that their solution is constantly trained on new invoices.
What type of companies provide invoice capture solutions?
Established account payable tech companies
These companies were to first to provide invoice data extraction solutions. Since their solutions were the first solutions on the market, some solutions are dated and rely on templates.
Amazon AWS Textract is a new comer in the field. Amazon also brings the ability to combine Textract with other services like ground truth. For example, ground truth could provide human validators to check documents that Textract can not process with a high level of confidence. This combination of services could allow companies to completely outsource their document processing. Such combined services can also be built on top of other companies’ solutions as well since most invoice capture solutions support APIs.
Startups leverage machine learning to build flexible solutions. Since the increasing commercialization of AI in the last 10 years, there has been an increase in application of AI into extracting structured data from semi-structured data. Outsiders could see startups as doomed after Amazon’s entry to the business. However, startups still have major advantages when compared to Amazon:
Hypatos, one of the startups in this space, pulled a UiPath. As you may remember, UiPath was the first RPA company to introduce a free version of their product in 2016. 3 years down the line, they are the most valuable RPA company with a latest valuation of ~7 billion as of April 2019.
Hypatos introduced a free version of their tool called Community Edition in November 2019. Though the free version produces lower accuracy products than their paid product, Subscription Edition, it could still be good enough for most use cases.
What is the complete list of companies that provide invoice capture solutions?
Below, you can find our initial list on the topic. Later we expanded our research and we are now keeping a regularly updated list of invoice capture vendors.
|Company||Number of employees on linkedin||Area of focus||Pricing||Largest customers||On prem solution||Type of solution|
|Amazon AWS Textract||N/A||Document data extraction||$0.05 per page**||Roche||Possible with AWS Outposts***||Pre-trained ML|
|Coupa||1000+||B2B spend management||Template based|
|Datamolino||11-50||Bookkeeping automation||Not template based|
|Docparser||1-5||Document data extraction||$0.05 per document (up to 5 pages per document)||SMEs||N/A||Template based|
|Docucharm||1-5||Document data extraction||N/A||Continuously trained ML|
|Hypatos||11-50||Document data extraction & advanced processing||Community Edition is free||PwC|
|Available||Continuously trained ML|
|Instabase||11-50||Document data extraction|
|pdfdata.io||1-5||Document data extraction||Template based|
|Proactis||501-1000||B2B spend management||Numerous Fortune 500||Available|
|SapphireOne||1-5||ERP, CRM, DMS and Business Accounting Software||Template based|
|Tabula (open source)||Not applicable||Table extraction||Template based|
|Tipalti||100-500||B2B spend management|
|Xtracta||11-50||Document data extraction||Available|
** Including key value pair+table extraction at a volume of 1M+ pages/month
*** Outposts was announced in AWS re:Invent 2018 but is not yet available. Post launch, services like RDS, ECS, EKS, SageMaker, EMR are announced to be the first services to be available
How to choose your invoice capture vendor?
Choose a provider that can provide a consistent data structure regardless of the text on the documents. There are two ways that deep learning based invoice capture companies work. Companies like Textract return key value pairs. So for example, if an invoice calls the total amount as “Gross amount”, the other calls it “Total amount” and another German invoice calls it “Summe”, Textract gives you the data in 3 different structures for these 3 documents. In one, you have a key value pair with the key “Gross amount”, in another “Total amount” and in the German one, you get “Summe”. Other providers like Hypatos designed consistent data structures that work for all invoices. In all 3 scenarios, you would get “Total amount” which the key they use in their output file. This makes analytics and processing easier as you don’t need to deal with many different structured data formats.
Ask for the false positive and manual data extraction rates. Then run a Proof of Concept (PoC) project to see the actual rates on the invoices received by your company.
- False positives are invoices that are auto-processed but have errors in data extraction. These are difficult to identify and can disrupt operations. For example, incorrect extraction of payment amounts would be problematic. Minimizing this should be the absolute focus.
- Manual data extraction is necessary when automated data extraction system has limited confidence in its result. This could be due to a different invoice format, poor image quality or a misprint by the supplier. This is also important to minimize but there’s a trade-off between false positives and manual data extraction. Having more manual data extraction can be preferable to having false positives.
We have not yet completed our benchmarking exercise but Hypatos’ benchmarking indicates that they lead the industry in both having minimal false positives and a low rate of required manual data extraction. This is the first quantitative benchmarking we have seen in this space and will follow a similar methodology to prepare our own benchmarking.
Leverage a PoC to measure the potential automation rate. This depends on the number of fields you expect to capture from the documents. A typical set of ~10 fields including items like purchase order ID, vendor name, vendor name etc. can enable data entry into ERP and payments. Best practice vendors achieve ~80% STP by extracting all of these ~10 fields with almost no errors ~80% of the time. Though there may be errors from time to time, manually checking the largest payments can ensure that no significant wrong payment slips through the net.
Ask for advanced processing options provided by the vendor. Extraction is the first step in data collection, it needs to be followed by data processing in most cases. For example, invoices need to be checked for VAT compliance (e.g. domestic invoices without VAT need to explain why VAT is excluded) and failure to do so could result in significant fines for the company depending on the country. Hypatos provides numerous advanced processing options, however we have not seen other vendors provide such features as they focus exclusively on data extraction.
Ask for how the solution learns about new invoices. Best solutions have an interface for allowing your team to help guide the solution. As your company’s employee picks the key-value pairs, the invoice capture solution takes note so it can be more confident about a similar invoice next time.
Evaluate the ease-of-use of their manual data entry solution. It will be used by your company’s back-office personnel as they manually process invoices that can not be automatically processed with confidence.
Beyond this, best practice procurement questions make sense. For example:
- How widely adopted is their solution? Do they have Fortune 500 customers?
- Are their customers happy with their solution and support? Could be good to ask an acquaintance from a company that is already using their solution. Since invoice automation is not a solution that would improve marketing or sales of a company, even competitors could share with one another their view of invoice automation solutions.
- What are the options to integrate the solution to your company’s systems (e.g. ERP)? Is IT on-board with the integration approach?
- What is their Total Cost of Ownership (TCO)? Different solutions use different units of pricing (e.g. price per page or price per document) which makes this comparison difficult. However, using a sample from your archives, you could have an estimate of the cost.
If you have more questions, feel free to contact us of course:
How can we do better?
Your feedback is valuable. We will do our best to improve our work based on it.