Not All Automations Are Equal: Understanding the Different Levels of Automation

When people hear the word automation, they often imagine a perfectly autonomous process that runs forever without human involvement.

In reality, very few business automations fit that description.

One of the biggest mistakes organisations make is viewing automation as a binary choice: either something is automated, or it is not. The truth is that automation exists on a spectrum. Different automations require different levels of oversight, maintenance, and human involvement.

Understanding these differences is critical when evaluating automation opportunities, estimating return on investment, and setting realistic expectations for ongoing support.

The Automation Spectrum

At MinuteView, we view automation as existing across four distinct levels of autonomy.

1. Autonomous Automation

Autonomous automations are the closest thing to a true "set and forget" solution.

These automations operate within highly controlled environments where the inputs, outputs, and business rules are well understood and unlikely to change. Once deployed, they can often run for months or years with little or no intervention.

Examples include:

• Synchronising data between two systems using stable APIs
• Automatically generating PDFs from approved documents
• Moving files between controlled repositories
• Updating metadata based on predefined business rules

In these scenarios, the same input produces the same output every time. The automation is highly predictable and exceptionally reliable.

The maintenance burden is extremely low, making these automations ideal candidates for long-term operational efficiency.

2. Managed Automation

Managed automation is where most business value is created.

These automations perform the overwhelming majority of work without human involvement, but occasional exceptions can occur. The key difference is that humans are not part of the process itself—they are only involved when an exception arises.

Consider a document submission workflow:

• Thousands of forms may be submitted every month
• 99% are processed automatically
• A small number fail due to unexpected data, formatting issues, or unusual edge cases

Rather than requiring human approval every time, the automation attempts to complete the task automatically and only raises an exception when something unexpected occurs.

Examples include:

• Form processing
• Document registration workflows
• Email-driven business processes
• Metadata extraction
• Automated drawing management
• OCR and document classification

The reality is that no organisation can predict every possible user input or every unusual scenario.

A user may enter unexpected characters into a form.

A supplier may upload a file with an unusual naming convention.

A document may contain data that was never encountered during testing.

These edge cases are not failures of automation. They are a normal part of operating real-world business systems.

The goal is not perfection.

The goal is to automate 95–99% of the workload and handle exceptions efficiently.

When an automation saves ten minutes per transaction across thousands of transactions, the occasional exception becomes insignificant compared to the overall business benefit.

This is why managed automation often delivers the highest return on investment.

3. Assisted Automation

Assisted automations are highly valuable but depend on systems that are outside your control.

A common example is browser automation or robotic process automation (RPA).

These workflows may:

• Log into websites
• Download reports
• Transfer information between systems
• Complete repetitive administrative tasks

The automation itself may work perfectly. The challenge is that the environment around it changes.

A website may be redesigned.

A login process may change.

A password may expire.

A multi-factor authentication requirement may be introduced.

In these situations, the automation requires occasional maintenance to remain operational.

Examples include:

• Browser automation
• Screen scraping
• Desktop application automation
• Legacy system integrations
• Systems without supported APIs

These automations still provide enormous value.

If an automation saves hundreds of hours every month but requires one hour of maintenance, the business case remains extremely compelling.

The important distinction is that these automations should be planned, supported, and maintained accordingly.

4. Human-Gated Automation

Human-gated automations intentionally include human involvement as part of the process.

Unlike managed automation, the workflow pauses and waits for a person to make a decision before continuing.

Examples include:

• Document approvals
• Financial approvals
• Change requests
• Quality assurance sign-offs
• Engineering review processes

In these cases, the human is not handling an exception.

The human is an intentional part of the workflow design.

This approach is often necessary where governance, compliance, risk management, or accountability requirements exist.

Why This Matters

Understanding the type of automation being implemented helps organisations answer several important questions:

Who owns the automation?

An autonomous workflow may require very little oversight.

A managed workflow requires someone to review exceptions.

An assisted workflow requires periodic maintenance.

Understanding ownership from the outset prevents confusion later.

Is automation financially worthwhile?

Not every process should be automated.

If an automation saves ten minutes per month but requires two hours of maintenance, the business case may not exist.

Conversely, an automation that saves hundreds of hours while requiring occasional maintenance is often an excellent investment.

What support model is required?

Different automation types require different support expectations.

Autonomous automations may need little more than monitoring.

Managed automations benefit from exception reporting and audit logs.

Assisted automations require operational ownership and periodic review.

How should success be measured?

Many organisations incorrectly measure automation success by asking:

"Does it work 100% of the time?"

A better question is:

"How much work does it eliminate?"

An automation that successfully handles 95% of transactions may still deliver extraordinary value if it removes thousands of hours of manual effort each year.

Don't Let Perfect Get in the Way of Better

One of the most common objections to automation is the belief that every possible scenario must be handled before a workflow can be deployed.

In practice, this approach often prevents organisations from realising any value at all.

The reality is that business processes are messy.

Users behave unexpectedly.

Systems change.

New requirements emerge.

Edge cases will always exist.

The objective of automation is not to eliminate every exception.

The objective is to eliminate as much repetitive work as possible while providing visibility, auditability, and control when exceptions occur.

Even when an automation requires occasional oversight, the productivity gains can be transformative.

The organisations that gain the greatest advantage from automation are not those that wait for perfection.

They are the organisations that understand the different levels of automation, apply the right approach to the right problem, and continuously improve over time.