Webbit
How to Choose the First Process to Automate (A Practical Checklist)
Not every process deserves automation. A simple scoring checklist for picking a first workflow that pays off, and the traps that waste time and budget.
Almost every team we talk to has a long list of things they would like to automate. The list is rarely the problem. The problem is choosing the first item on it, because that first project sets the tone. If it is small, clear and useful, the team gains confidence and momentum. If it is vague, risky or politically complicated, it stalls and the whole idea of automation gets a bad name.
This guide gives you a simple way to choose. You can run it in an hour with a whiteboard or a spreadsheet, and you don't need any technical background.
Start with the work, not the technology
It's tempting to begin with a tool: a chatbot, a voice agent, a new CRM. Begin instead with a plain description of the work. Ask the people who do it, not only the people who manage it, and write down three things for each task: what triggers it, what the person actually does, and what "done" looks like.
Good candidates tend to share a few traits. They happen often, they follow rules you can describe, they move information between tools, and they make people sigh when you mention them.
The five-question score
Score each candidate from 1 (weak) to 3 (strong) on the questions below, then add up the points. You're looking for the candidates near the top, not for a perfect number.
| Question | What a high score looks like |
|---|
| How often does it happen? | Daily or many times a week, so a small saving repeats. |
| Can you write the rules down? | A new colleague could follow a one-page instruction and get it right. |
| How few exceptions are there? | Most cases look the same, and unusual ones are easy to spot. |
| Are the systems reachable? | The tools involved have an export, an integration or an API. |
| What if it goes wrong? | A mistake is cheap to catch and fix, not a legal or financial incident. |
A process that scores well on all five is usually a safe first project. A process that scores low on rules or risk isn't a bad idea, just a later one.
A worked example (illustrative)
Imagine a small property services company comparing three ideas. The scores below are made up to show the method. They aren't results from any real client.
| Candidate | Frequency | Rules | Exceptions | Systems | Risk | Total |
|---|
| Copy web-form enquiries into the CRM and notify an owner | 3 | 3 | 3 | 3 | 3 | 15 |
| Reply to customer complaints | 2 | 1 | 1 | 3 | 1 | 8 |
| Build the weekly sales report | 2 | 3 | 2 | 2 | 3 | 12 |
The enquiry routing wins clearly. It happens constantly, the rules are obvious and a mistake is easy to catch. The complaint replies score low because every case is different and a wrong answer is costly. Complaints can be automated later. For now a person should draft and decide, and the first project should be the one that builds trust.
Rule of thumb. If you can't explain the process to a new hire in ten minutes, you aren't ready to automate it. Simplify it first, then automate.
Traps that waste the first project
- Choosing the most impressive idea. A flashy demo rarely matches the work your team repeats every day.
- Automating a broken process. If the steps are confusing, a machine will repeat the confusion faster. Remove unnecessary steps first.
- No owner. Every automated workflow needs a named person who watches it, handles exceptions and decides when to change it.
- Starting with the hardest exceptions. Handle the common path first and send everything unusual to a person with the context attached.
- Skipping the baseline. Note how long the task takes today, so you can tell afterwards whether the change helped.
What a good first project looks like
- One trigger, such as a form submission, an email or a status change.
- A handful of steps that follow written rules.
- Two or three tools that need to share information.
- A human review point for anything unusual.
- A clear measure of success, such as hours saved or enquiries answered the same day.
Projects with this shape can often be scoped in days rather than months, which keeps the risk low and the learning fast. Once the first workflow is live and trusted, the next one gets easier, because you already have access, naming conventions and a routine for reviewing exceptions.
Signs a process isn't ready yet
Some processes are better left alone for now. Watch for these warning signs.
- The rules change constantly. If the policy is rewritten every month, the automation will be out of date as soon as it's built.
- Nobody agrees how it works. Two people doing the same task in different ways means the process needs defining before it needs automating.
- It happens rarely. A task done twice a year is cheap to do by hand, even if it is annoying.
- The inputs live on paper or in someone's head. If the information is never written down digitally, there is nothing for a system to read.
- Mistakes are expensive and hard to spot. Keep a person in charge until the checks are strong.
Run the exercise this week
- List the work (30 minutes). Ask each team member for the three tasks they would happily never do again.
- Describe each one (30 minutes). Note the trigger, the steps, the tools used, who is involved and how often it happens.
- Score them (15 minutes). Use the five questions above, and discuss any score where people disagree.
- Choose one or two (15 minutes). Prefer the highest score that has a willing owner.
- Write the baseline. Time three real examples so you can compare afterwards.
Bring the shortlist and the baseline to a conversation with your automation partner. It turns a vague wish into a concrete brief.
Want a second pair of eyes?
In an automation audit we do this scoring with you. You describe the process, the systems and how often it runs, and we review the fit, flag risks and suggest a realistic first scope. If a process isn't ready for automation, we'll say so and explain what would need to change. For many teams the right starting point is business process automation across two or three connected tools.
Want to put a number on the work first? Read the real cost of manual work and use the worksheet there.
Frequently asked questions
How many processes should we automate at once?
Start with one. A single workflow that works and is trusted teaches you more than five half-finished ones. Add the next once the first is stable and someone owns it.
Do we need to replace our existing software?
Usually not. Most first projects connect the tools you already use, so information moves between them without retyping. Replacing software is a separate decision with a different cost.
What if our process has lots of exceptions?
Automate the common path and route exceptions to a person with the details already attached. Over time you can learn which exceptions repeat and handle those too.
Home · Blog · Book an Automation Audit