Four signs a business process is ready to automate
Focus first on one or two workflows that show at least one of these signs. Each sign comes with a customer example, with results as each case study states them.
Sign 1: Changes take too long
When a change that should take days takes months, the process has outgrown its system. WIJEindhoven, a foundation with 450 employees whose work depends on legislation that keeps changing, said its old control system was not flexible enough to follow those changes and was also very expensive. According to its case study, it built a new system in less than seven months and now manages it itself, so new concepts can be realized quickly.
Sign 2: Maintenance keeps eating the budget
Aging systems carry costs that never show up on a project plan: slow changes, large teams and overhead just to keep them running. According to its case study, FacilityXL had an aging system that delayed development, carried high maintenance costs and took months to deliver new features. After rebuilding its online portal, the case study reports development ten times faster than traditional methods, build time reduced by more than 50 percent and no more costly user-based licensing. As managing director Mark Houtbraken put it, “An extremely short time-to-market gives us a big competitive advantage.”
Sign 3: Growth is straining the process
A process that works at one scale can fail at the next. IVC Evidensia runs a network of 2,500 clinics in 20 countries, and as it grew, handling more than 500 employee transitions a year by hand brought delays and errors, according to its case study. The case study reports that those transitions are now automated, that about 1,200 bot tasks run each month, and that errors fell by 80%.
Sign 4: People spend their days on manual effort
Manual effort hides in spreadsheets, re-keying and email handoffs. Asana’s 2023 Anatomy of Work Global Index, a survey of more than 9,600 knowledge workers in six countries, found that people spend 58% of their day on work about work, meaning coordination rather than skilled, strategic work. According to its case study, Arval had a legacy system that needed extensive manual effort for inspection, damage assessment and settlement. Processing time for returned vehicles then fell from 22 days to 2, and 50% of returns are now handled without human involvement.
A readiness test for any candidate process
Run every candidate through the same six questions before you commit. A candidate does not have to pass them all, but a no to the second question means mapping the process comes before automating it.
| Question | Ready when | Not ready yet |
|---|---|---|
| Does it repeat? | Ready whenIt runs often, with the same steps each time | Not ready yetIt is rare, or each case is built from scratch |
| Can the owner describe it? | Ready whenThey can map every step, decision and handoff | Not ready yetNobody can say how it really runs |
| Do the rules hold? | Ready whenDecisions follow criteria you can write down | Not ready yetDecisions depend on judgment that changes case by case |
| Does it cost something visible? | Ready whenIt is slow, error-prone or expensive, and people feel it | Not ready yetIt already meets its targets |
| Is a mistake bounded? | Ready whenA wrong result is caught and fixed without real operational risk | Not ready yetA wrong result spreads into other systems before anyone notices |
| Can the systems be reached? | Ready whenThe systems involved offer an API, a database or a file export | Not ready yetThe work lives only on paper or in people’s heads |
When to fix a process before you automate it
Automation makes a process faster, not better. Three situations call for fixing the process first.
- The people who run it disagree about how it works. Automation will only run the disagreement faster. Map the process with them, agree on one version, then automate that.
- It changes every few weeks. Automating a moving target means rebuilding it each time. Stabilize the rules first, or choose a platform where business teams can change the workflow themselves.
- It already meets its targets. Automating a process that works adds a project and a dependency without removing a cost. Start where the pain is.
Sort the exceptions before you choose a tool
Most processes have a main path and a tail of exceptions, and the tail decides which tool fits. Take a sample of recent cases, mark every one where a person had to step in, and ask whether the reason can be written as a rule.
- Workflow
Exceptions that follow a rule belong in the workflow. A rule-based workflow handles them predictably and is straightforward to test. Chapter 5 compares workflows with bots. - Governed AI
Exceptions that need judgment are where governed AI fits. Reading an unstructured email is an example. The agent works inside the workflow, within limits you set. Chapter 6 covers where agentic AI replaces fixed rules. - Bot
A stable, single task on a system with no API is the case for a bot. Chapter 5 sets out when a bot is enough.
Gartner’s June 2025 release offers a similar split: use AI agents when decisions are needed, automation for routine workflows and assistants for basic retrieval.
Start with one process and the people who run it
Pick one process. Map every step, decision and handoff before choosing a tool, and bring in the people who run the work every day. Their early feedback decides whether the result works in practice as well as on paper. Then set success metrics tied to business outcomes, such as fewer manual interventions, lower maintenance cost or faster changes, and compare the results with the cost of leaving the process as it is.
Ownership matters too. In WEM No-Code, business teams own workflow design, logic and day-to-day management, while IT grants access to external data sources and handles security sign-off, according to the WEM No-Code business process automation page. The page also says single workflows reach production in weeks and larger applications take months.