The majority of automation projects in South Africa do not fail on the shop floor, they fail in the boardroom, kickoff meeting, or first discovery workshop, long before any workflow is ever made live.
The problem is simple: we all want to quickly determine what we want to automate, without ever stopping to think about why we want to do it, whom we want to do it for, and how we do it today.
Mistake 1: No Clear Business Problem
Many automation projects kick off with “We need AI” or “We must automate this platform.”
Instead, we should be starting with what we call a clear business problem.
If you cannot complete the sentence “We are doing this to reduce or increase [metric] by [X] percent,” the project is not ready for workflow design.
How to fix it
Every automation initiative should be anchored by 1-3 clear, measurable outcomes, such as:
- Reduce onboarding turnaround from 10 days to 3
- Cut manual claims rework by 30 percent
These outcomes become the guiding principle for scope, design and prioritisation.
Mistake 2: Automating Fantasy Instead of Reality
Slide decks usually portray smooth, linear processes. Real-world processes, however, are far more complex in many African organisations. Approvals might happen via WhatsApp, while workarounds exist because the system does not fully support the process.
This mistake means that the automated process is sure to fail when users attempt to use it.
How to fix it
Spend time getting a better understanding of the real process.
This means shadowing users, studying real-world examples, and studying actual data. Map the current state, including workarounds, to get a better understanding of the process.
Mistake 3: No Process Owner
Automation projects are frequently owned by IT or the external vendor, especially when they are technology-centric projects.
The problem, however, is that the business process belongs to the business process, not IT. This mistake means that the process will not evolve after the project has gone live.
How to fix it
This means nominating a process owner who is a business process. This could be an operations manager, a customer services manager, a claims manager, etc.
Their job is to ensure that the process continues evolving after the project has gone live.
Mistake 4: Ignoring Change Management
Most people don’t resist automation because they enjoy doing things manually. They resist change because they don’t understand what’s changing, how it affects their role, or how they’ll be supported through the change.
How to fix it
Recognize automation as a change management activity, not just a technical one. Communicate early, show the difference between before and after, run pilot programs, and deliver simple playbooks and training inside the tools that teams are already using. Help the individual understand the value of automation for their own performance.
Mistake 5: Going Too Big Too Soon
Big end-to-end automation projects look good on paper, but they fail when they hit the complexities of the business.
How to fix it
Automate one or two high-value workflows, such as onboarding, case management, or collections, in one business unit. Show the value first, learn the patterns that worked, and then extend automation gradually through the organization using the play, pause, and repeat approach.
Building Automation That Actually Works
When these upstream issues are solved, automation is no longer an experiment with risk; it is an operating discipline that an organization can depend on to operate at scale, manage risk, and liberate teams to focus on the work that really needs human judgment.