What to Automate First: An ROI-Based Way to Prioritize

TL;DR: The best first automation is rarely the exciting one. It's the boring, high-volume process that's quietly bleeding time and money, the one everyone's stopped noticing because "that's just how it works." Score your candidate workflows on three things: how often it runs, how expensive its errors are, and how manual it is today. Automate the highest score first. Getting the first one right matters more than the automation itself, because it funds and proves everything after it.

Why most automation projects pick the wrong thing first

Teams tend to automate what's visible, the process someone's currently frustrated with, rather than what's expensive. Or they start with something big and ambitious, which takes months, drains goodwill, and delivers nothing for a long time.

The result: a stalled project, a skeptical team, and a "we tried automation and it didn't work" story that makes the next attempt harder. The fix is to pick a first workflow that's high-return, contained, and fast to prove.

The scoring framework

Rate each candidate workflow 1-5 on three factors, then multiply:

• Volume: how often does this run? A process that happens 500 times a month has far more to gain than one that happens twice.
• Error cost: what does a single mistake cost, in penalties, rework, or lost margin? Small errors × high volume is the classic hidden drain (see procurement).
• Manual effort: how much human time and re-keying does it take today? The more people are the glue between systems, the bigger the win.

Score = Volume × Error cost × Manual effort. The highest score is almost always where you should start, and it's usually not the process anyone would have named first.

A quick worksheet

For each candidate, answer:

1. How many times does this run per week or month?
2. When it goes wrong, what does it cost, and how often does it go wrong?
3. How many systems does someone re-key data into to complete it?
4. How many people touch it, and how often does it stall waiting on a hand-off?
5. Could we ship an improvement in weeks, not quarters?

The process that scores high on 1-4 and passes 5 is your first automation.

Red flags: what NOT to automate first

• A broken process. Automating a bad workflow just lets you make mistakes faster, redesign it first. Automation amplifies whatever it's pointed at.
• A rare, high-variance process. Low volume means low return, and high variance means it's hard to automate reliably.
• Something with no clear owner. If nobody owns the outcome, nobody will adopt or maintain the automation.
• The most complex thing you've got. Save it for after you've banked an early win and built trust.

How we prioritize with clients

The scoring is the start; the discipline is in the discovery. We map how work actually flows, find where it breaks, and pick the first workflow to redesign based on return, not noise, then prove it before expanding. That's step one of our method The 5-Step Process We Use to Redesign a Team's Workflows.

Frequently asked questions

How do I decide which process to automate first?

Score each candidate on volume (how often it runs), error cost (what mistakes cost × how often), and manual effort (how much re-keying and human time it takes). Multiply the three and start with the highest score, usually a high-volume, error-prone, heavily manual process, not the flashiest one.

What's a good first automation project?

A contained, high-volume, error-prone workflow you can improve in weeks rather than quarters, often invoice matching, PO processing, or a compliance checklist. It should have a clear owner and a measurable before/after so the win is undeniable.

Should I automate a process that's currently broken?

No, fix the process first. Automation amplifies whatever it's pointed at, so automating a broken workflow just produces errors faster and at scale. Redesign the process, then automate it.

How long should the first automation take?

Aim for weeks, not months. The goal of the first project is to deliver a clear, provable win that builds trust and funds the next one, not to boil the ocean with a company-wide platform rollout.

---

Not sure which workflow will pay back fastest? Schedule a process audit and we'll score your candidates with you, and start with the one that pays for itself.