SMB automation often gets confused with "installing an ERP" or "digitizing the company." That level of ambition is rarely needed upfront. Three concrete workflows, one by one, pay for themselves in under a quarter and free up someone on the team for work that actually requires thinking. This article describes which ones, with data, cost, and when it does NOT pay off.
Gartner puts 20–30% of a small business back-office's time on repetitive administrative work (Gartner, 2024). Every hour reassigned to sales or retention is worth more than its direct cost. That's where automation of concrete processes — not "digital transformation" — produces visible returns.
Client onboarding end to end
The most common case. A web form with the data, automatic validation, creation of the record in the management system, contract generation pre-filled with data, delivery via Signaturit or Docusign, sign-up in the CRM and notification to the sales team. Everything with no human touch other than validating the submission.
Average manual time for a full sign-up runs 15–25 minutes depending on sector. With the automated flow, it drops to under 30 seconds of human intervention. In a business with 40 sign-ups per month that's 10–15 hours recovered from a single person. Typical tools: n8n or Make, from €20/month depending on volume.
Recurring payments and dunning
Stripe, GoCardless or Redsys issue the charge. If it fails, the system triggers an automated notice to the customer. If the notice doesn't resolve it, it escalates to the commercial owner with full context on the debtor (history, average ticket, active campaigns). Nobody watching dashboards to catch missed payments.
Stripe itself publishes that 11% of recurring payments fail on the first attempt due to expired cards, credit limits or transient banking issues (Stripe Dunning Guide). A well-designed dunning workflow (smart retry + right-toned communication + escalation) recovers up to 70% of that 11%. In a SaaS business billing €40,000/month, that's about €3,000/month of recovered revenue.
A well-designed dunning workflow recovers up to 70% of the payments that fail on first attempt.
Actionable operational alerts
Notifications when something requires concrete human action: order pending picking for more than 2 hours, stock below minimum, invoice overdue by X days, production incident above a threshold. Sent to the right channel (Slack, WhatsApp Business, email) and to the right person by type. No more "just in case" periodic checks.
WhatsApp Business Platform charges from €0.03 per utility conversation initiated by the business (Meta for Developers). For 300 alerts a month that's under €10, and it frees the team from 30–40 minutes daily checking the system.
When NOT to automate
Three clear cases:
- Processes that run fewer than 5 times a month. Building the workflow costs more than running it by hand for a whole year.
- Processes that change a lot. If the flow gets rewired every month, redesigning it costs more than executing it.
- Processes with high cost of error under no supervision. Large payments, legal decisions, negotiations. Automation needs a human checkpoint in these cases.
How to calculate the return
Useful formula: (executions per month × minutes per execution × team hourly cost) minus (monthly tool cost + monthly amortization of the workflow build). If the flow pays back in under 12 months, automate. If it takes longer, rethink or drop it.
In some of the cases we've done, the return shows up by the second month. In others, when volume is low, keeping the manual process is more profitable — wait until the business grows.
