Workflow automation examples for small businesses

Good workflow automation examples for a small business are lead routing, invoice capture, customer and staff onboarding, review requests, stock alerts, payment reminders and a weekly report email. Each one has a clear trigger, a fixed set of steps and an owner who checks the exceptions. Most of them are plain rules; only work that starts from messy input, such as a PDF bill or a free-text email, needs an AI step.

Our glossary entry on workflow automation explains the terms. Below are nine examples by department.

Nine examples at a glance

"Workflow tool" means n8n, Zapier, Make or similar; all three can run every example here.

Sales and marketing

1. Lead routing. A form is submitted on the website or a new number messages you on WhatsApp. The workflow checks the CRM for an existing contact so you do not create a duplicate, assigns a rep by region or product, sends the rep an alert with the enquiry, and creates a follow-up task. If nobody has replied within an hour, it tells the sales manager. This is rules. An AI step helps only when the enquiry is free text and you want it tagged by product or urgency before it is assigned.

2. Review requests. An order is marked delivered, or a clinic visit is closed. The workflow waits two days, then sends a short message with a link to your Google Business Profile. Send it to every customer, not only the happy ones: Google's policy for Maps says businesses must not "selectively solicit positive reviews" and must not offer incentives for reviews. Plain rules.

Finance

3. Invoice capture. Supplier bills arrive at a bills@ address as PDFs or phone photos. The workflow saves the attachment, runs OCR, and a language model pulls out the supplier, invoice number, date, net amount and tax. A rule checks that the lines add up to the total. Anything that fails the check, or that the model is unsure about, goes to a person in a review queue before it reaches the accounts. This one needs AI, and it needs a human in the loop step. We have built an OCR pipeline of this kind, for invoices, IDs and forms, with human review.

4. Payment reminders. Every morning a schedule reads the list of unpaid invoices. Three days after the due date it sends a polite reminder with the invoice number, amount and a payment link; at ten days a firmer one; after that it alerts the owner. Razorpay can send a payment_link.paid webhook when a link is paid, so the workflow marks the invoice paid and stops the reminders. Rules only. Keep the decision to chase a valued customer by phone with a person.

Operations, support and HR

5. Stock alerts. Shopify sends an inventory_levels/update webhook whenever a stock level changes. The workflow compares the new level with your reorder point and, if it is below, messages the buyer with the item, supplier and usual order quantity, or drafts a purchase order for approval. Rules.

6. Customer onboarding. A deal is marked won in the CRM. The workflow creates the project, a shared folder from a template and the first tasks, sends the welcome email with what happens next, and books an internal handover from sales to delivery. Rules.

7. Support triage. An email lands in the support inbox. A language model classifies it (billing, delivery, fault, complaint), sets the urgency and drafts a reply from your help articles. A person reads and sends the draft. Complaints and anything mentioning a refund go straight to a manager. This needs AI, and the human check stays until you have measured how often the draft is right.

8. New starter onboarding. An offer is accepted. The workflow creates the email account, adds the person to the right groups and shared drives, asks IT for a laptop, sends the manager a day-one checklist, and schedules check-ins at one week and one month. Rules.

9. Weekly report email. Every Monday at 08:00 the workflow pulls last week's sales, new leads, ad spend and overdue invoices, works out the totals and the change on the week before, and emails a one-page summary to the owners. Rules. You can add a model to write two sentences of commentary, but the numbers must come from the data, never from the model.

Which examples need AI, and which are plain rules?

Seven of the nine are rules: if this happens, do that, with conditions you can write down in advance. Two need AI, invoice capture and support triage, because the input is unstructured: a scanned bill or an email written in someone's own words.

A useful test: if you can write the decision as a short list of conditions, use rules. If a person has to read something to decide, an AI step can do the reading, and a person should check its output until you know its error rate. Our AI workflow automation projects follow that split.

How to pick your first one

Choose the example that runs most often, follows the clearest rules and has one person who will own it when it fails.

Then put numbers on it. Our AI automation ROI calculator works out hours freed per month from hours per week, people and the share automated, and subtracts running costs. If the workflow will carry customer or financial data and run thousands of times a month, a self-hosted build through our n8n integrations service keeps the data on a server you control.

If you have a list of repetitive jobs and want to know which one to automate first, send it to us and we will say plainly which are worth building. Book a consultation.