
- Membership
- Certification
- Resources
- Events
- Community
- About
- Help
When I first started really digging into AP automation, I thought the goal was pretty straightforward: reduce manual work, process invoices faster, improve accuracy and make everyone's life easier. Simple, right? Well, automation apparently did not get the memo that it was supposed to be simple.
Our journey has taught me that AP automation is rarely just a technology project. It quickly becomes a process project, a data project, a people project, a change management project and, occasionally, a “Why on earth is the system doing THAT?” project. Honestly, that may be the biggest thing I wish I had known when we started.
One of my biggest lessons was realizing that you cannot automate your way around a messy process. When we started looking closely at AP, there were manual processes, workarounds and different ways of handling exceptions that had developed over time. Some made perfect sense. Some probably made sense five years ago. And some fell squarely into my favorite category: “Well, that's the way we've always done it.”
Before pushing harder on automation, we needed to understand the processes themselves. Who owns this? Why are we doing this step? Does anyone actually use this information? Why are we entering the same information twice? Why are we tracking something outside the ERP? Could the system already do this? Those questions became just as important as choosing the technology. That's when I learned my first big lesson: If you automate a bad process, congratulations. You now have a faster bad process.
Another surprise was discovering how important it is to understand the technology you already own before going shopping for something new. And as we dug deeper, we found opportunities within our existing environment that we had not been fully utilizing. That changed the way I looked at automation. Instead of immediately asking, “What new solution should we buy?” I started asking, “What are we already paying for that we're not using?”
Sometimes automation doesn't require another huge implementation. It may be better configuration, a workflow change, automated coding or a smaller automation built around an existing process. We have also found opportunities to use tools like Microsoft Power Automate to address specific AP pain points. I love technology, but I also love not spending money unnecessarily.
I've also completely changed the way I think automation solutions should be evaluated. If I could change one thing about the typical vendor demonstration, I would say, "Please stop showing me the perfect invoice.”
Give me the ugly one.
Give me the invoice with multiple lines, freight, questionable coding, a PO issue and some strange attachment nobody understands. Show me what happens when the supplier doesn't match, or the information isn't exactly where the technology expects it to be. Clean transactions are not where AP spends most of its time. Exceptions are.
Today, one of my first questions would be, “Show me your exception process.” Then I want to know who fixes the exception, how AP knows something failed, what stays manual and how much work is created on the other side. A process can look incredibly automated in a demonstration, but it doesn't feel very automated if AP spends the rest of the day cleaning up the exceptions.
Then there is the data. Automation has a funny way of uncovering every inconsistency you've been living with for years. Supplier records, coding, payment methods, duplicate information, old workarounds and, of course, the spreadsheet that somehow became a critical business system even though nobody remembers who created it.
Automation depends on good information. When the underlying data isn't clean or standardized, technology doesn't magically fix it. It just processes what you gave it, very efficiently. That’s why I would tell anyone beginning an automation journey to make data cleanup and process standardization part of the automation project from the beginning, rather than treating them as side projects you'll get around to later.
Trust me. You'll get around to them.
The automation will make sure of it.
Another thing I have learned is that AP needs a real voice in these projects. IT understands technology. Finance understands financial objectives. Procurement understands purchasing. Consultants understand their solutions. But AP knows what happens at 4:45 on a Friday afternoon when an invoice doesn't match the PO, the approver is unavailable and someone has now marked their third email URGENT.
That operational knowledge matters.
AP should be involved when requirements are developed, solutions are evaluated, processes are redesigned and implementation decisions are made. We shouldn't just be the people receiving the solution after everyone else has designed it. We're the people who have to live in it.
The people side of automation has probably been one of the most important lessons for me as a leader. Leadership hears the word “automation” and thinks efficiency, scalability and savings. Employees can hear the exact same word and think, “Are you automating my job?”
You cannot ignore that.
I've become much more intentional about explaining why we are automating. The goal isn't to eliminate the value AP employees provide. It's quite the opposite.
I don't want talented people spending their careers typing the same information into the same fields over and over again. I want them managing exceptions, analyzing problems, working with suppliers, improving processes and strengthening controls. But leaders have to help employees understand what their role looks like on the other side of automation.
I've also learned that an email announcing a change is not training. People need walkthroughs. They need demonstrations. They need to understand why the process is changing and what they are expected to do differently. Most importantly, they need the opportunity to ask questions, especially the questions that begin with, “Okay, but what happens when...?”
Those are usually the good ones.
There is another reality we don't always talk about when discussing automation: Technology fails.
A workflow can fail. Something may not route correctly. Data can create an exception. A process may behave differently than expected. Automating something does not eliminate the need for controls. In some ways, it creates an entirely new control question: How do we know when the automation didn't work?
That question has become increasingly important to me. Automated processes still need exception monitoring, clear ownership and a way to identify failures. Otherwise, everyone assumes the system handled something when it actually didn't. For me, success isn't simply asking, “How much did we automate?” I also want to ask, “How much rework did the automation create?” That can tell you a very different story.
If I were starting an AP automation journey tomorrow, I would not begin with vendor demonstrations. I would begin with AP. I would map the real process, including the ugly exceptions. I would look at the data. I would understand the technology we already have. I would establish baseline metrics and ask the AP team where they are spending their time and what frustrates them the most.
Then I would bring in the technology.
And when vendors show up for demonstrations, I would ask the uncomfortable questions: What happens when the automation fails? How are exceptions handled? What requires customization? What isn't included? How does it integrate with the ERP? What will AP still be doing manually afterward? What happens after implementation? And what is the actual total cost once everything is included?
Most importantly, I wouldn't automate something simply because we can. I would automate it because doing so makes the overall process better.
What are you waiting for?