Assessment
How to Choose Which Business Process to Automate First
Last updated
The short answer
Automate first the process that combines high volume, a lot of manual handling time, a stable set of steps, digital inputs and systems you can connect to, and a clear owner. Score your candidate processes on value (hours, errors and delay it costs today) and feasibility (stability, data, systems and exceptions), then pick the one that scores well on both. Avoid starting with the most complex or most political process, even if it is the most painful: the first project's job is to prove the method and build trust.
Step 1: List candidates and measure them
Ask each team for the two or three tasks that take the most time or cause the most rework. For each, collect four numbers: monthly volume, minutes per item, people involved and error or rework rate. Estimates are fine at this stage, but write them down; they become the baseline.
Step 2: Score value and feasibility
| Criterion | Score high when | Score low when |
|---|---|---|
| Time cost | Hundreds of hours a month across the team | A few hours a month |
| Error cost | Mistakes reach customers, payroll or the accounts | Errors are caught and cheap to fix |
| Stability | The steps are the same for everyone and rarely change | The process is being redesigned or varies by person |
| Inputs | Data arrives digitally, in systems, forms or email | Inputs are paper or phone calls |
| System access | Systems have APIs or supported connectors | Legacy systems with no access except the screen |
| Exceptions | Most items follow the normal path | Most items need judgement or escalation |
| Ownership | One named owner can make decisions | Several teams share it and disagree |
Step 3: Pick the first project
- Shortlist the processes that score well on both value and feasibility. A high-value process with poor feasibility is a second or third project, after the groundwork.
- Prefer a process whose result is easy to see, such as days saved in the close or a queue that clears faster.
- Check that someone can review exceptions during the pilot.
- Agree the success measure and the decision you will make if it is hit or missed.
Common first choices
Processes that often score well are invoice processing, bank and ledger reconciliation, employee onboarding across HR and IT systems, data subject access requests, contract intake and review, and moving data between a CRM and an ERP. Each is high-volume, rule-heavy and spread across several systems.
Frequently Asked Questions
- Should we automate the most painful process first?
- Only if it also scores well on feasibility. The most painful process is often painful because it is unstable, poorly owned or full of exceptions, which also makes it the hardest to automate. Proving the method on a feasible process first makes the harder one easier to fund and deliver.
- How long should a first automation project take?
- A single, well-chosen process can usually be assessed in days and built and proven in a few weeks. If the first project is planned to take many months, the scope is probably too broad.
- What if no process scores well on feasibility?
- Then the first project is groundwork: standardise the steps of the most valuable process, get data access to its systems, or name an owner. That is a better use of budget than automating a process that will change next quarter.
Keep exploring
Related reading
- Process AssessmentOne workflow mapped, measured and costed, with a fixed implementation proposal.
- AI readiness assessmentWhat an assessment covers and a 12-question self-check.
- ROI calculatorsEstimate what automating a process is worth with your own numbers.
- Automation SprintRedesign and build one workflow, then prove the economics.
Next step
Get a measured answer for your workflow
We map one process, measure what it costs today, and give you a fixed proposal.