Joakim Kyllönen • Sun Oct 02 2022
Companies often struggle with turning purchase invoice automation numbers into operational efficiency. This effect seems to persist regardless of company size, location, or industry.
For the past three years, we have helped accounting offices of various sizes systematically develop the automation of their ongoing bookkeeping processes. We’ve noticed three key issues that should be addressed when building automation solutions.
Opening, reading and figuring out what to do with an invoice is one of the most time-consuming steps of manual purchase invoice processing, and it often accounts for more than 50% of the time used. A single manually completed action requires that the human opens and reads the invoice.
Half of time savings will be lost unless every task on an invoice is automated
It is important to understand and track the degree of automation for every process that is completed manually. The end goal should be to maximize the number of invoices where every manual step is automated.
Exceptions that are difficult to automate (comments from invoice approver, credit invoices, mismatching sums, etc.) make up 1-10% of purchase invoices and are therefore often excluded from the automation scope. However, exceptions can become problematic if one has to check every single invoice to find that rare one in a hundred exceptions.
An exception on 1% of invoices may require that all 100% of invoices are opened, in order to find that one exception case.
Checking every invoice for an exception will incur the cost of reading an invoice for every invoice, even if every other action is automated.
Ultimately, the responsibility of any process will fall upon a human rather than a machine. An individual responsible for the correctness of the purchase ledger who has little incentive for efficiency will want to control every decision the automation makes, incurring a high cost in terms of time saved. The degree to which this is happening is difficult to measure, but we believe that clarity and simplicity in the process are crucial to minimizing this effect.
The process should have a clear handover point between machine and human with information on what actions/decisions still have to be made or controlled.
The best solution, in my opinion, is to transfer the responsibility of invoice automation to a dedicated automation team/individual. It is then up to this party to identify which invoices are automated with sufficient confidence and exclude these from the tasklist of those responsible for manual invoice processing.
A human spends a lot of time opening and interpreting invoices when processing purchase invoices manually. The primary purpose of purchase invoice automation should be to minimize the number of times this happens. To ensure that you maximize the amount of time saved by your automation efforts, you should ask yourself three questions: