The first automation project in a business does more than save time. It decides whether the team trusts the next one. Pick a process that is too ambitious, too rare or too political, and automation becomes "that thing that did not work". Pick well, and the next project gets funded on the strength of the first.
Here is the method we use with clients to choose.
Step 1: List the candidates
Ask each team lead a single question: "What task do your people do over and over that involves copying information from one place to another?" You will typically get ten to twenty answers. Write each down with the team, the systems involved and roughly how often it happens.
Step 2: Score each one
Score each candidate from 1 to 5 on five factors:
- Volume: how often it happens. Daily, high-count tasks score high; quarterly tasks score low.
- Rule clarity: how consistently it is done. If two people would handle the same case the same way, it scores high.
- Pain: how much time, frustration or error cost it creates today.
- Data readiness: whether the inputs are digital and the systems involved are accessible through an API, import or at least email.
- Low risk: how contained the damage is if something goes wrong. A misrouted email is low risk; a wrong payment is not.
Add the scores. Candidates above 18 are strong first projects. Anything with a 1 on data readiness or low risk should wait, however attractive it looks otherwise.
Step 3: Sense-check the winner
Before committing, check three things:
- There is an owner. Someone who runs the process today, wants it improved and can make decisions.
- Success is measurable. You can count the volume and time today, so you can show the difference later.
- It is contained. It touches two or three systems, not seven. Bigger processes can come later, once the infrastructure and trust exist.
Good first projects we see often
- Supplier invoice entry in finance teams with hundreds of invoices a month. See invoice processing automation.
- Customer order entry from email and PDF at distributors and manufacturers. See purchase order automation.
- Shared inbox triage for orders@, ap@ or support@ addresses. See shared inbox automation.
- Proof-of-delivery processing at carriers and 3PLs, which also speeds up billing. See shipping document automation.
- Weekly reporting that someone builds by hand from several exports. See reporting automation.
A worked example: scoring five candidates
Here is how the scoring might look for a mid-sized distributor. The numbers are illustrative.
| Candidate | Volume | Rule clarity | Pain | Data readiness | Low risk | Total |
|---|---|---|---|---|---|---|
| Customer order entry from PDF | 5 | 4 | 5 | 4 | 4 | 22 |
| Supplier invoice entry | 4 | 4 | 4 | 4 | 3 | 19 |
| Weekly sales report | 3 | 5 | 3 | 4 | 5 | 20 |
| Credit applications | 2 | 3 | 3 | 3 | 3 | 14 |
| Year-end stock count | 1 | 3 | 5 | 2 | 3 | 14 |
Order entry wins: it happens all day, the rules are clear, mistakes are costly and the inputs are digital. The weekly report scores well too and is low risk, which makes it a good second project or a quick win to run in parallel. The stock count is painful but rare, so it should wait however much the team dislikes it.
Common mistakes when choosing
- Picking the loudest complaint. The most frustrating task is not always the most valuable one. Score it like everything else.
- Picking the biggest process. End-to-end order-to-cash is a programme, not a first project. Pick one step of it.
- Ignoring the people who do the work. They know the exceptions. Involve them early and they become the project's strongest supporters.
- Skipping the baseline. Without a measure of today, nobody can prove the improvement later.
Processes to avoid as a first project
- Anything that happens rarely, however painful, such as annual audits or year-end.
- Processes that are being redesigned anyway, or where the team disagrees on how they should work.
- Tasks that depend on judgment that has never been written down.
- Anything that moves money without a person approving it.
Then measure
Record the baseline before you start: volume per week, average handling time, error or rework rate. Measure the same things a month after go-live. That before-and-after comparison is what earns the second project.
Our ROI calculator helps put a number on each candidate, and an automation assessment does this scoring with you, on real data, in about a week.