You can’t build an effective purchase invoice automation with traditional methods

Almost all medium and large organizations have been struggling with a manual purchase invoice process for years. I have met with hundreds of large organizations over the past couple of years and have become familiar with their purchase invoicing processes. The more I have seen different organizations, the more convinced I have become that the purchase invoice process cannot be automated using traditional methods.

Purchase invoice automation has gotten stuck to a low level

In most companies, the level of automation of non-PO purchase invoices has remained unfathomably low. In general, at least 80% of purchase invoices are posted manually by a person and only about 20% are posted automatically. In addition, on average, at least 70% of invoices are sent manually for circulation to reviewers and approvers.

Manual posting and routing of one purchase invoice take several minutes and causes much unnecessary administrative work and frustration. The high manual workload can be seen in the fact that the quality of accounting for purchase invoices is often poor and the invoices are in circulation within the organization for a long time. Especially the turn of the month is usually a time when the manual work of purchase invoices causes annoyance.

Does such a large amount of manual work make any sense in 2022?

In my opinion: no, it makes no sense whatsoever.

The reason for low automation is not companies’ laziness or inability to build automation rules. The reason is rule-based automation itself. Rule-based automation does not lend itself to high levels of automation, no matter what kind of guru is building the rules.

The representatives of the companies I meet often feel inferior and think that everyone else’s purchase invoice process is more automatic and smoother than theirs. Let me assure you, that this is by no means the case, regardless of which purchase invoice system is in use.

Why can’t the purchase invoice process be automated using traditional methods?

If you think about your situation regarding the automation of the purchase invoice process, you will most likely recognize most of the things on the list below. At least I hear these every day when I talk to different organizations.

• There are very few repeated purchase invoices that are similar in proportion to the entire invoice volume – 20% maximum
• Many different products and services are purchased from the same suppliers, so supplier-specific invoicing rules do not work
• Suppliers forget to put reference numbers on invoices and they put them in the wrong fields
• When scanning, the reference numbers remain on the image of the invoice and are not transferred to the XML file that is read by the purchase invoice system
• There are so many companies that building rules for each company would be such a massive job that the invoices are rather be processed manually
• It would be possible to use the automation rules, but no one has time to familiarize themselves with them – in addition, they must be maintained if you want them to work for the long run
• We used to have one person who made a huge amount of rules for us, but the person left and no one has any idea how those rules work
• We were promised a high level of purchase invoice automation, but we were left to build the automation alone and our hands are full with other work

These are just some of the things I constantly hear when I talk to customers. I believe that these situations may be familiar to you as well.

Artificial intelligence changes the way how purchase invoices are automated

Artificial intelligence changes the entire playing field of purchase invoice automation because AI builds the automation rules automatically, and there is no need to manually build and maintain a traditional jungle of rules. AI learns from historical purchase invoices by reading how the invoices have been posted and how they have been reviewed and circulated in the organization. This is how AI automatically learns the accounting rules directly from the existing data, without a person having to define complex rules. In addition, AI is capable of creating much more complex rules than a human being could ever do.

AI has been able to raise the automation level of purchase invoices from 20% to over 90%. The degree of automation increases as soon as AI is put into use, and its ability increases the most during the first 6 months of use.

In addition, AI uses all the available data on purchase invoices (both written and image) to create automation rules. Thus, the automation rule is not based on any single reference number, but on all the information on the invoice. This makes the operation of automation significantly easier, as it reduces the need to communicate with suppliers.

If you want traditional rule-based automation to work, you have to get both your organization and suppliers involved in the process, so that invoices always have the exact right reference information to trigger the automation rule. However, often the challenge is that the invoice data is incomplete and no matter what kind of rules are built, they will not work without the missing data.

If you still feel bad about the fact that your organization has not reached a 90% level of purchase invoice automation, I recommend that you should finally relax about this issue.

Discover what kind of opportunities AI brings to purchase invoice automation and take the first steps toward a new era for your organization. Remember that you don’t need to buy purchase invoice AI without first testing its capabilities with your purchase invoice material. Snowfox.AI can give you a free AI ability test which will give you precise information on how well AI would process your purchase invoices automatically. In addition, we will give you a calculation of how much the use of AI would cost you.