"What should we automate first?" is the most useful question a team can ask, and the one most often skipped. Teams tend to start with whatever is most annoying, not with what will pay back fastest.
Here's the scoring we use in discovery workshops. It takes about an hour with the people who actually do the work.
1. List the repetitive work
Write down every task that happens on a schedule or every time something arrives: a new lead, an invoice, a support email, a form submission. For each one, note how often it happens and roughly how long it takes.
2. Score each workflow
- Volume: how many times a week does it happen?
- Time: how many minutes does a person spend each time?
- Rules: can you explain the steps to a new hire on one page?
- Inputs: does the information arrive in a predictable place, such as email, a form or a CRM?
- Cost of mistakes: what happens if it's late or wrong?
High volume, clear rules and predictable inputs make a strong first candidate. AI helps most where the inputs are messy, such as free-text emails, PDFs or call notes, because it can read and classify them before the rules take over.
3. Estimate the payoff honestly
Multiply hours saved per week by the cost of those hours, then add the value of speed: leads answered in minutes instead of hours, invoices chased the day they're due. Subtract the build cost and the running cost of the tools and AI models.
Be conservative. If a workflow only pays back under optimistic assumptions, it isn't your first project.
4. Keep a human in the loop where it matters
The best automations don't remove people. They remove the copy-paste. Anything with real risk, such as refunds, contracts or messages to important clients, should go to a person for approval with everything prepared for them.
5. Start small, measure, then expand
Automate one workflow end to end, using the tools you already have. Measure it for a few weeks: time saved, error rate, response times. Then move to the next one. This is how automation earns trust inside a team.
If you'd like help scoring your workflows, that's exactly what our discovery call is for.
