Before You Automate: Workflow Automation & ERP Implementation Risks
It was an ERP implementation, more than fifteen years ago. The vendor team were sharp, experienced, and working hard. The client's retail supply chain was complex, and the allocations process had everyone turned around. The team had tried multiple configuration options. The client kept coming back frustrated. The conversations were going in circles.
After one particularly long session, I asked the head of supply chain a simple question: “Can you draw what a pack looks like for these people?”
She picked up a marker and drew it. And in that moment, three days of circular conversations were resolved.
Because what the vendor team had been building toward was not what the client actually meant. Same word. Completely different picture.
Why unclear system language cause automation failure
In retail operations, the word “pack” carries a lot of weight.
It can mean a consumer unit: socks sold in a three-pack. It can mean a ratio pack: a small, medium, or large allocation sent to a store as an initial inventory drop. It can mean a case pack: twelve bottles, twenty-four cans, the way a customer buys a product in bulk.
The same word, completely different commercial concepts.
The ERP system had its own interpretation of ‘pack’. It also had something it called bundles and something called kits.
The client's requirement sat somewhere between all three, and it didn't map cleanly to any of them. Until that gap had a name, every configuration built on top of it was an efficient way of encoding the wrong answer.
That is not a technology problem. It is a language and comprehension problem. And it’s extraordinarily common in ERP implementation services and retail implementation projects.
Why this matters
Manual processes have a built-in correction mechanism: people. When a picking team member notices something looks off, they ask.
When a buyer sees a price that doesn't make sense, they pick up the phone. Human judgment operates as a constant, low-level audit of what the system is producing.
Retail automation removes that friction, which is the point. But it also removes those small correction moments.
If the logic underpinning your automation is based on a misunderstanding of how the system interprets a barcode, a naming convention, or a unit of measure, the automation will execute that misunderstanding perfectly, every single time, at whatever volume you've scaled it to.
Decimal points are a good example of this.
A system configured to four decimal places talking to a system configured to two creates rounding errors that compound across thousands of transactions.
You won't see it in the moment. You'll find it in a stock count three months later, in a stock discrepancy, or in a supplier reconciliation that nobody can explain.
Those same issues often sit behind inventory errors and the hidden causes of stock discrepancies that consume time later.
What workflow automation readiness actually requires
It is not enough to understand what your system can do. You need to understand what it means by the words it uses, and how those meanings carry across every integration point in your landscape.
Before you automate anything in a retail environment, map the following:
Terminology alignment: Does your business use the same word as the system? Does your ERP use the same word as your WMS, your POS, and your supplier portal? If not, which definition wins, and where?
Rules at the boundary: How does your system interpret a barcode when it crosses into another system? What happens to a SKU description when it moves from your master data into a third-party logistics platform? This is where retail integration support becomes critical.
Unit and precision mismatches: Currency decimal places, unit of measure conversions, date formats. These are unglamorous and they will break your workflow automation if you don't surface them before go-live.
The unmapped concept: What does your business need that the system doesn't have a clean concept for? That gap needs a decision, not a workaround.
These are foundational issues in retail inventory management, retail supply chain management, and broader supply chain management in retail industry settings.
When automation scales the wrong process
There's a version of this that plays out at scale. A business identifies a process to automate, scopes it, builds it, and launches it. It works. Then they add another automation. Then another. And nobody goes back to the first one.
Each new automation is built on the assumptions of the last. The conceptual misalignments that were tolerable in a manual environment become load-bearing walls in an automated one.
A 2024 HBR study on automation complexity found that, much like stress on the body, a small amount of AI assistance improves performance, but there is a tipping point where oversight load becomes so heavy that people can no longer effectively monitor what they're supervising. You end up with more automation in retail and less understanding of what it's actually doing.
That creates risk, slows retail operational efficiency, and can lead to disruption in retail when systems fail at scale.
The governance question most retail leaders aren't asking is not ‘does the automation work?’ It’s ‘what breaks in six months when the complexity builds?’
Formal review cycles, not just at go-live but three, six, and twelve weeks post-implementation, are how you catch the drift before it becomes a crisis and avoid an ERP implementation failure.
Your pre-automation readiness check
Before any workflow is built, get your functional, technical, and commercial people together and ask:
Can everyone here define the key data objects this automation will touch, and do those definitions match without prompting?
Where does data cross a system boundary in this workflow, and who owns the translation rules at that point?
Which human is accountable for the output of this automation, and do they understand it well enough to know when something is wrong?
When will you return to review this in the context of everything else added since?
Whether guided internally or by a trusted retail consultant, these questions matter.
“Deep system awareness is not about being technical. It’s about refusing to automate something you don't yet fully understand.”
TL;DR: What retail leaders need to know
The hidden mechanism: Automation amplifies the logic it's built on. If that logic contains a misunderstood data concept, a boundary translation error, or an unmapped business requirement, the automation doesn't just fail. It scales the failure.
Diagnostic Questions:
Can your project team define every key data object in this automation the same way, without prompting each other?
Do you know what happens to your data at every system boundary in this workflow?
When did you last review your earliest automations in the context of everything built since?
Decision Framework: If you cannot clearly explain what the system means by a term, and confirm that the meaning is consistent across every system it touches, you are not ready to automate that process.
That discipline supports business process automation in retail, stronger retail business automation, automated retail systems, and ultimately improved retail performance. The conversation you avoid now is the problem you manage later.
Retail improvement, made practical.
Leadership thinking that drives change.
Sign up to receive new articles and strategic guidance.
