NEW WHITE PAPER: The Great Invoice Scanning Myth

whitepaper icon18th February 2014

After 25 years of working within the document management industry I have seen many changing challenges faced by businesses. Throughout this time a consistent challenge faced by organisations surrounds the processing of invoices and I’d like to share some of my thoughts on this issue.

Why bother?

Only the fittest survive and without exception all successful companies are continually looking to improve their bottom line. Whether that’s improving their core product, reducing overheads, streamlining staff or investing in more productive machinery, they’re always striving to increase profit.

A subject close to many accountants hearts is the time spent receiving and processing purchase invoices. This can be a thankless task, especially if the volumes are high and a large proportion of invoices require intervention or approval. So, it’s not surprising finance teams are keen to rationalise this process and release valuable time for more useful and profitable work.

Help – we need an automated invoice scanning and approval process!

Here at Document Genetics, we’re often approached with requests by financial directors to solve their supplier invoice problems.

Well, here’s the first bit of free advice – there is no magic bullet; the more automation you require, the more the solution will cost and the greater the complexity of implementation.

Unfortunately, there’s now a perception that low cost solutions exist which will allow you to magically scan a batch of paper invoices and automatically extract via OCR (Optical Character Recognition) the information you need (e.g. invoice number, order number, date, supplier name, VAT total, net total etc.). If you want this technology, it costs, both in software and more importantly professional services and training cost.

Going a stage further, if you want line item recognition (i.e. extracting product codes, quantity, unit costs etc.), this is a whole new level of sophistication and cost.

The Invoice Scanning Myth

Don’t believe everything you see or hear.

There are specialist companies and even software authors, promising the earth. All have their preferred technology and clever OCR algorithms which claim to be superior to the competition. They can successfully demonstrate their state of the art system with their “pet sample” of invoices. Not surprisingly, the results will be astonishing.

However, I’m afraid there’s often a lot of “smoke and mirrors” here! It may have taken weeks or even months to tweak a system to cope with their “pet samples”, so it stands to reason the results will be good!

A much better test is to throw a real world sample at them – the ones with missing purchase orders, those printed on multi-part paper by a dot matrix printer, strange paper types or sizes.

Hand on my heart; I’ve never seen a successful implementation of line level recognition of invoices. I’m sure they’re out there but they are few and far between. However, I have seen lots of awful ones. This quote is typical:

“If I had my time again, we wouldn’t have gone down this route. We’ve never achieved the recognition levels we were promised and the vast majority of invoices still require human intervention. The return on investment just hasn’t stacked up”

We’re sometimes approached by companies processing 500 to 1500 invoices a month wanting a system which will seamlessly capture all the information they need with virtually no human intervention. The truth is at this monthly volume, you’ll never get a realistic return on investment and you’d be better taking a phased approach which initially addresses the digitising, indexing and routing of invoices for approval.

How do the bureaus do it?

Here’s a final thought – how do the outsourcing bureaus process large volumes on a daily basis. Surely, they must use the most sophisticated OCR software solutions? In actual fact, many bureaus key from image (i.e. they use a human to manually type the index fields from the invoice) because it’s bullet proof with a very low overhead in terms of support and maintenance. We’re not saying OCR technology doesn’t work, but you must have the volumes, the budget and the commitment to implement the system properly.

How can Document Genetics help you?

Our aim is to provide genuine advice and realistic guidance – warts and all. We’re not going to make incredulous claims with unbelievable ROIs - and if that loses us customers, then so be it.

Firstly, we need to understand what your problem areas are. Is it the sheer volume of invoices, is it matching invoices with GRNs and delivery notes or is it approval of invoices throughout the business? Secondly, how do you receive invoices – is it a combination of paper and electronic formats and if so what are the proportions of each (e.g. 65% paper, 35% electronic)? What ways do you accept or receive electronic invoices – PDFs via email or perhaps XML via FTP? Thirdly, do the bulk of your invoices come from your top 20 suppliers (i.e. can we apply the Pareto analysis or 80/20 rule?)? All these questions have a bearing on the solution we can recommend, both in terms of software and implementation cost.

Consider your volumes, as this has a direct bearing on the possible time savings and the final return on investment. Below is a simple chart which illustrates the level of sophistication of the solution, the typical cost and the recommended monthly volumes for this type of system. We’ve had to make some assumptions here but it’s a good place to start.

Scanner Interface

Barcode Recognition

Auto import of electronic documents

Full page OCR

Key from image

Header level field extraction

Header and line level field extraction

Workflow Approval

3 way matching*

Typical solution cost

Recommended volumes per month




£6k to £9k





£9k to £15k

500 to 10,000


£15k to £30k

5,000 to 30,000

£30k to £75k

> 30,000

What are your volumes and budget and where do you fit into this chart? Is your expectation of functionality in line with your budget and monthly volumes of invoices?

Accounts Payable (AP) Automation Checklist

Here’s a checklist of useful questions to ask prior to engaging in any supplier invoice approval, accounts payable automation or invoice scanning project.

  • What are your monthly volumes of purchase invoices?
  • What proportion are stock and non-stock invoices?
  • What proportion are paper vs. electronic?
  • What are the electronic formats and methods of receipt?
  • Do you have a top 20 suppliers, who raise the vast majority of invoices (80/20 rule)?
  • What fields do you collect off your invoices?
  • What’s your reason for wanting to automate your supplier invoice process?
  • What proportion of invoices matches your purchase orders, GRNs and delivery notes and requires no intervention?
  • How do you current resolve matching issues?
  • Do you manually approve invoices through the organisation?
  • Do you have an existing document management system?

Contact Document Genetics

If you’d like to discuss your supplier invoice application with Document Genetics, we’ll be delighted to help.

Author: Joe Hyde – Director at Document Genetics

Tel:         +44 (0) 1604 671177

Email: This email address is being protected from spambots. You need JavaScript enabled to view it.     Web: