Automating a business process: the complete SMB checklist (2026)

Many SMB owners want to automate but do not know where to start. Which processes are actually suitable? What do you need on paper up front? Which questions do you ask a partner so you do not end up six months later stuck with a tool you cannot leave? This article is a checklist you can tick off yourself. No consultancy needed, no intake call required. For the build itself we often look at document processing or sales automation. Just a workable structure we use ourselves with every new client, with the criteria, questions and pitfalls that make the difference between a process running in production after three months and one that stalls in week six.

Which processes are suitable for automation?

Not every process belongs under an automation. Some are too variable, too creative, or simply pay back too little to earn back the build cost. Our rule of thumb: a process is a good candidate if you can tick off five criteria.

1. Repetitive. The process happens at least a few times per week, ideally several times per day.
2. Rule-based. You can describe the steps in if-then-else logic. Work that depends on feel is out.
3. Digital. The data already sits in systems, emails, forms or spreadsheets.
4. Measurable. You know how much time it takes and how often it happens.
5. Not creative. There is no free human judgement involved. Writing a tailored proposal stays human work.

Ticking four or five of these? Then there is a strong case underneath. Three or fewer? Pick a different process. For the broader context first, read our AI automation SMB guide.

Which information should you document up front?

The biggest cause of failed automation projects is not technology, it is poor documentation at the front end. Before you call a partner, capture these seven points.

Point	What to capture	Example
1. Trigger	What starts the process	New email to info@, new form submitted
2. Steps	What happens in order	Read email, classify, enter in CRM, send confirmation
3. Decision points	Where the flow branches	Existing customer yes/no, amount above €5,000
4. Systems	Which tools the process touches	Gmail, HubSpot, Exact, Slack
5. Frequency	How often per day or week	20 times per day, 100 times per week
6. Volume	Number of items per action	1 email = 1 lead, 1 email = 5 invoices
7. KPIs	What you want to improve	Lead time from 2 hours down to 5 minutes

How do you carefully estimate ROI?

We published a detailed calculation method in calculating workflow automation ROI. The short version is a formula with three variables.

• Time saved per month = (current time per action in minutes) × (number of actions per month) × (estimated automation percentage 70-90 percent).
• Monetary value per month = (time saved in hours) × (internal hourly rate, often €40 to €75 for SMBs).
• Payback period in months = (one-time build cost) / (monthly value minus tooling cost).

According to McKinsey research on SMB automation, the realistic average sits between 60 and 80 percent of routine tasks, not 100.

Which questions do you ask a partner?

1. Ownership. “Do we get the full code, configuration and documentation in our hands?”
2. Price. “Is this a fixed price or an estimate?”
3. Hosting. “Where does the solution run and who pays for hosting?”
4. Security. “Which data leaves our environment and which providers do you use?”
5. Support. “What is the agreement after go-live?”
6. Notice period. “How do we move on if we want to continue elsewhere?”
7. References. “Can we talk to two existing clients in our sector?”
8. Guarantee. “What if it works differently than promised after go-live?” Good answer: a correction guarantee within a reasonable window (with us, 30 days, see terms and conditions).

Which pitfalls should you check up front?

1. No measurable baseline. Prevent by tracking for three days how much time the current process actually takes.
2. First case too big. Prevent by starting with a process that costs 5 to 15 hours per month.
3. Using AI where rules would do. Prevent by first checking whether an if-then-else flow can handle it.
4. No human in the loop for risky steps. Prevent by always building an approval step for outbound communication and financial actions.
5. No exit strategy. Prevent by agreeing from day one on code ownership and exportability.

What does a good automation brief look like?

An automation brief is a document of at most two A4 pages that you send to a partner before a first call. A good brief contains six parts.

1. Context. A few sentences about your company, your sector, and why you are automating now.
2. The process. Describe in at most half an A4 what currently happens. Use the documentation checklist.
3. Goals and KPIs. What do you want to achieve concretely? Per KPI a baseline and a target.
4. Constraints. What are your hard limits? Budget, data location, sector compliance.
5. Non-goals. Equally important. What is explicitly not in scope?
6. Timeline. When do you want to be live? Realistic: six to eight weeks for a first flow.

How do you measure success after go-live?

• Time saved per week. Realistic target: 60 to 85 percent reduction.
• Error reduction. A good flow has fewer than 5 percent exceptions that need manual work.
• Lead time. From trigger to completion. A process that used to take 24 hours and now takes 15 minutes is not a beauty contest, it is direct revenue or customer satisfaction gain.

What do suitable processes look like per sector?

Theory is nice, examples are more useful. Below are four short cases per sector with the kind of numbers we see come back in intakes. Compare them to your own working day and you will likely spot a candidate of your own within two minutes.

• Law firms: intake and case classification. An 8-lawyer firm received 40 to 60 emails per day on the central intake address. The secretariat spent 90 minutes per day classifying and forwarding them. After automation: classification within 2 minutes per email, lead time down from 4 hours to 12 minutes, the secretariat gets 7 hours per week back for case work.
• Accountancy: invoice intake and posting. An SMB accountant processed 800 invoices per month for 35 clients, manual re-keying took 18 hours per week. After document-extraction automation: 600 invoices fully automatic, 200 with a human in the loop for exceptions, time spent drops to 4 hours per week, error rate from 6 to 1 percent.
• Recruitment: candidate intake and pre-screening. A recruiter received 120 applications per week, read each CV in 4 minutes, and did 8 hours of screening on Monday and Tuesday. After automation with scoring logic on hard criteria: 70 percent auto-rejected or auto-progressed, the recruiter reviews 36 CVs by hand, screen time drops from 8 to 2.5 hours per week.
• E-commerce: order processing and after-sales. A webshop with 220 orders per day had three staff dealing with order confirmations, delivery questions and return requests. After automating standard statuses and the returns flow: 65 percent of tickets are handled automatically, one staff member takes the full after-sales package, the other two move to product expansion.

Which failure patterns can you spot early?

Half of the automation projects that stall did so not on technology but on signals that were already visible in week two. Four patterns we keep seeing, and how to pick them up before they cost money.

Failure pattern	Early signal	What you do then
Scope keeps growing	Every weekly demo brings new wishes, no cut decisions	Pause the build, plan a re-alignment, write down what stays in scope and what moves to phase 2
No internal owner	Demo questions come from a different person each time, nobody can say yes until delivery	Assign one decision maker with a mandate, set a fixed weekly demo slot
Too many exceptions	By week 3, 40 percent of cases fall outside the original rules	Split the flow into a main flow plus a human-in-the-loop track for exceptions
Unclear baseline	No one knows what the current process really costs, ROI discussion becomes guesswork	Pause and do three days of time tracking after all, recalculate the business case

A good partner surfaces these signals themselves in weeks two and three. If all you get in that window is status updates without critical questions, the partner is in execute-the-brief mode, not in think-along mode. Name that out loud before it is too late.

Who needs to approve what internally?

An automation project touches more departments than an owner usually thinks at the start. A sales flow touches marketing, finance and operations. A document flow touches finance, IT and privacy. The mapping below helps you write down who signs off on what before kickoff.

1. Ultimately responsible (1 person). Usually the director or an operational lead. Signs off on scope, budget and go-live.
2. Process expert (1 or 2 people). The staff who run the process daily. Provide the documentation, judge acceptance tests, flag exceptions.
3. IT or data owner. Approves the systems being connected, the credentials, and where data is allowed to go. At bigger companies a separate role, at smaller ones often the director.
4. Finance control. Approves the TCO, payback period, and tooling cost. Especially relevant if you pay monthly for a no-code platform.
5. Privacy or compliance control. For processes involving personal or client data. In SMBs often via the external DPO or the external accountant.

How do you start concretely?

1. Step 1: identify your top 3 processes with the five criteria. Pick the most painful one with the most repetition.
2. Step 2: document in two hours. Fill in the seven-point checklist for the chosen process. Half an A4 is enough.
3. Step 3: request two quotes. Ask the eight questions. Compare the answers before you look at the price.

Want to do this together? Plan a free intake call. In thirty minutes we walk through your three candidates. Prefer to run the numbers yourself first? Read calculating workflow automation ROI, or get the broader context in our AI automation SMB guide.