How it helps and how it hinders.
We can’t talk about bookkeeping and accounting without talking about technology.
Something I’ve noticed is that technology can separate us from the fundamentals, and can make the foundational concepts of bookkeeping and accounting harder to learn and grasp. It’s like the difference between listening to a podcast on a topic, or researching and writing an essay on a topic. It’s in the doing that we solidify our understanding.
Technology can also separate us from the data. When there’s a lot of data to be processed, that can be a help, but if we don’t know what it is doing, chaos can ensue. I’m not against automation. I’m against badly implemented automation.
You cannot oversee a system if you don’t understand how it works. And now, with AI being such a prevalent part of our technological life, this is more important than ever.
I have a lot of experience, thoughts, and opinions on technology in the accounting industry, and that is what this page is dedicated to.
There is one singular reason that QuickBooks Online (QBO) and other cloud softwares like Xero are much harder to learn than traditional desktop softwares. It is because they fuse two completely different methodologies of bookkeeping together and pretend that’s normal.
All bookkeeping software falls into one of two camps. Entry-centric bookkeeping, where you enter a transaction from the paperwork. Or feed-centric bookkeeping, where you enter a transaction from the bank feed. Sage is entry-centric. Xero leans feed-centric.
QBO does both. It pushes bank statement bookkeeping, categorizing transactions from the feed, which is cash accounting in practice. And at the same time it’s a fully robust accrual software, with bills, bill payments, invoices, invoice payments and combined deposits.
Having both options actually super charges the software, if you understand what is going on. You can use the accrual tools for detailed and operational bookkeeping, and use the bank feed to automate repeated entries that need no source document, and the bank feed can also speed up the bank reconciliations. When used for accrual source document based bookkeeping, the bank feed is the beginning of the last step in the bookkeeping workflow. It is that first step of the bank reconciliation that marks an existing entry as cleared.
Except that because it does both, it is harder to learn and understand. It is fusing two bookkeeping methodologies together, and nobody is saying this explicitly out loud. Because it can be used for either accrual bookkeeping or cash accounting, new users come at it from only one of those two perspectives and the software clashes. If you are doing accrual source document based bookkeeping, then starting with the bank feed is like starting at the end and working backwards. If you are doing cash accounting, then entering invoices and client payments and then re-adding deposits directly to sales from the bank feed causes the very common double entered income problem.
Intuit marketing tells you that QBO is easy to use and learn, which is part of the problem. No one tells you AutoCAD is easy to learn, because it isn’t. But that doesn’t make it less powerful.
So if you’ve sat down with QBO and felt like you were missing something obvious, you weren’t. It is confusing. It’s not you. It’s them.
To deeply understand QuickBooks Online, you need to deeply understand what that bank feed is actually doing. That’s what my QBO Bank Feed Masterclass is for. If you want the core of it in 40 minutes first, there is also the QBO Bank Feed Mini Course.
Over the years I have seen the shift in how bank feeds affected how bookkeeping gets done, and affected the reliability of the bookkeeping file.
Bank feeds are in practically all cloud bookkeeping apps today. I think they are a really informative precursor to what we’re about to start dealing with in this new AI era. AI may be new, but “smart categorization” is something that we in the accounting industry have had to contend with now for over a decade.
Effectively, the accounting profession has already been living with AI’s first draft, and we haven’t fully unpacked what that changed yet. Understanding that can not only shed light on how to make accounting and bookkeeping better today, but also position us to better approach how AI is going to affect things in the future.
The arrival of the bank feed has affected how people both learn and approach bookkeeping.
I’ve always been of the position that, as a bookkeeper for many years, I had the advantage of learning desktop bookkeeping. It was slower than cloud bookkeeping, as it didn’t have bank feeds, but it was more involved. Now, as I train new and learning bookkeepers, they are facing a much steeper learning curve with cloud bookkeeping; with all the bank feeds, integrations, online payment portals, all while trying to grasp the fundamentals of bookkeeping that I so easily take for granted.
The problem with experts is that we don’t remember not knowing. It’s not just facts for us. We feel it in our bones. Those of us who learned bookkeeping and accounting from desktop applications, or even a set of physical ledger books, gained an immersed experience, and we were able to adapt better to the bank feed technology and mess of online payments and transactions. We didn’t just know the theory. We knew context.
Remember the QuickBooks Desktop home screen? A beautiful customer flow of “Create Invoice” with an arrow pointing to “Receive Payments” with an arrow to “Record Deposits” with reconcile at the far right of the screen.
The bank reconciliation came at the END of the bookkeeping workflow.
But the bank feed puts the first step of that bank reconciliation, marking what has cleared the bank, at the beginning of the bookkeeping workflow.
Cloud bookkeeping softwares with a bank feed are blending two bookkeeping methodologies and just assuming everyone ‘gets’ it. We’re asking users to start at the end of the accrual bookkeeping workflow and work backwards, and then we wonder why they’re having a hard time.
Bookkeepers who learned on desktop had a bit of a hill to climb when they moved to the cloud. They got there, because they already understood the workflow underneath. Newer bookkeepers without that experience had a mountain. Not all of them made it to the top. Even students coming out of two year programs struggle with the bank feed, because it separates them from fundamentals they never had a chance to put into practice.
That bank feed tech is a help and a hinder. It’s helped make accrual bookkeeping quicker for those who understand accrual bookkeeping workflows, and we’ve all seen the chaos that ensues when it is used by those who don’t.
The problem is, the bank feed looks so straight forward, and is so very convincing when it’s wrong. But it’s such a powerful tool once you figure out what it’s doing. And you can only figure out what it’s doing if you deeply understand bookkeeping. Our frame of mind completely affects how we see and interpret and use the tool. How much we believe it. How much we know how to question it. That judgment has to come before the tool, not after. Using the tool before understanding makes learning harder.
I see the bank feed revolution as a precursor to AI. If I thought the bank feed separated my students from the fundamentals, AI is going to do that on steroids.
May I posit: because of the bank feed revolution, our accounting industry is actually one of the best positioned to understand and tackle this oncoming AI revolution. We’ve seen that learning to do by doing is the best way to learn, and productive struggle builds knowledge and confidence. And that a powerful confident looking tool like the bank feed can hallucinate and misguide users, especially when the people selling it tell us it’s easy to use and makes our life easier.
We have the advantage of experience this time. Let’s not waste it.
So, did bank feeds prepare us for AI, or warn us about it? It depends if we’re looking straight at it and intentionally preparing, or if we’re going to just drift and have the AI tides pull us along.
Remember computers? Not the box you’re reading this on, but the people who computed numbers before machines did that job. The women of NASA who helped us get to the moon. That role didn’t become obsolete when PCs came to market. You still needed people who understood the equations. They just used the PCs.
Same with paper and pencil ledger books. Excel replacing them meant fewer errors adding up long streams of numbers, because Excel did the adding. Tools make us quicker and more accurate.
But AI is different.
When you put 2 + 2 into a computer or into Excel, you can trust the answer 4 that it gives you. With AI, you cannot. Which means the professionals with experience will be the only ones able to use this tool to its strongest extent, because you need to understand what the tool is doing to use it properly. Otherwise you end up with convincing looking chaos.
And that brings us to the obvious question everyone in AI land is grappling with. If experience is what empowers professionals to use AI, and AI is doing the entry level junior bookkeeping roles that produced that experience, where do future bookkeepers get it?
My aha moment came from testing AI against my own bookkeeping simulation. I was able to train it to do the bookkeeping because I had the answer key. I knew where to check it, where to correct it, and where to tell it to stop doing things.
So here’s what I think the answer looks like: I call it “Parallel Bookkeeping.”
A junior bookkeeper does the file alongside the AI. The training runs both ways. The junior catches where the AI hallucinates, by comparing it against their own work. The AI catches the weird niche situations that usually take more experience to spot. Every time a firm onboards a new client, a junior does the parallel bookkeeping with AI to build the AI bookkeeping agent for that specific client, because every client is different and that’s where the weirdness lives.
The junior gets experience doing real bookkeeping, and experience reviewing files, which is the harder skill.
It doesn’t have to be limited to juniors. But it keeps an entry level position open, and it keeps experience in the loop, which is what we fear using AI will take away. Parallel Bookkeeping neatly solves that problem of losing the entry level positions that teach experience, while learning and using a powerful tool in a way that can support growth, not chaos.