20 Business Process Automation Examples (2026)

Business process automation means software runs a repeatable, rule-based process from a defined trigger to a defined output, with no person moving the data by hand. Examples of business process automation are invoice processing, AP three-way matching, purchase orders and expense approvals in finance; employee onboarding, offboarding, leave requests and hiring in HR; lead routing, lead scoring, nurture sequences and recurring reporting in sales and marketing ops; support ticket routing, escalations, IT ticket intake and access provisioning in service and IT; and inventory reorder, contract generation, e-signature routing and cross-tool data sync in operations. The 20 business process automation examples below are organised by department, and each one names the trigger, the systems involved and the point where a person still signs off.

What counts as business process automation?

A business process is a repeating sequence of steps that turns an input into an output: an invoice arrives and gets paid. Business process automation (BPA) hands that sequence to software so the steps run from the trigger without a person copying, forwarding or reminding, the same way every time, with a record.

The useful split is between rule-following work, which has a trigger you can name, inputs you can list and an outcome that follows from them (a bill that matches its purchase order within tolerance gets paid), and judgement-call work. Judgement-call work looks the same on the surface, but the decision is the hard part (which of two reasonable vendors to pick), and software that takes on that end usually goes by the label intelligent automation. This page catalogues the rule-following end; the judgement-call half is covered in AI automation examples.

The 20 business process automation examples at a glance

Finance processes you can take off your team’s plate

Finance processes have the clearest rules and the strictest audit expectations. Accounts receivable work such as invoice reminders and payment matching follows the same pattern as the AP examples. Month-end work is covered in month-end close automation.

1. Invoice processing

Before: an AP clerk re-keyed each vendor email and filed the PDF.

Now: a new email in the AP mailbox triggers extraction and creates a bill in QuickBooks, Xero or NetSuite with the PDF attached; a missing PO number or an unknown vendor goes to an exceptions queue. Sign-off: AP reviews the queue.

2. AP three-way matching

Before: someone compared the PO, receiving record and invoice across three screens.

Now: a posted bill is compared line by line with the PO and goods receipt; matches within tolerance go to the payment run, and mismatches are tagged with the reason (price variance, quantity short, no PO). Sign-off: the controller approves anything outside tolerance.

3. Purchase orders

Before: approval happened in chat, finance typed the PO, and nobody could find the approval later.

Now: a request form creates a draft PO in procurement, routes it by amount and cost centre, records the approval on the PO and emails the vendor the PO number. Sign-off: the budget owner approves above the company’s threshold.

4. Expense approvals

Before: receipts arrived as chat photos, and finance re-keyed them into payroll at month end.

Now: a receipt submitted in the expense tool is matched to the card feed, checked against policy (category, limit, receipt present) and queued for payroll. Sign-off: the manager approves out-of-policy items.

HR processes that run on triggers, not reminders

HR processes fail on reminders; each of these replaces the reminder with a trigger.

5. Employee onboarding

Before: the manager emailed IT, HR and the team lead separately, and logins arrived late.

Now: an offer marked signed creates the HRIS record, opens an IT ticket for the role’s standard kit, schedules the LMS training checklist and posts a welcome on the start date. Sign-off: the manager confirms the access list once.

6. Offboarding

Before: access removal depended on memory.

Now: a termination date in the HRIS schedules suspension in the identity provider, revokes app seats, opens a device return request and tells payroll the final pay date. Sign-off: IT confirms deprovisioning is complete.

7. Leave requests

Before: requests went to the manager by message, and payroll found out at month end.

Now: a request in the HRIS checks the balance, routes to the manager, blocks the calendar on approval and feeds payroll. Sign-off: the manager approves.

8. Hiring workflow

Before: the recruiter copied applications into a spreadsheet and chased interviewers.

Now: an application creates the candidate in the ATS, screening answers tag it, interview slots come from interviewers’ calendars, and each status change sends the matching email template. Sign-off: the recruiter advances or rejects.

See the jobs teams hand over →

Sales and marketing ops you stop touching by hand

Trigger-based capture for sales and marketing workflows lives in the CRM. HubSpot’s own documentation describes a workflow that can “enroll all contacts that have submitted a specific form, then send a marketing email and assign these contacts to a user”.

9. Lead routing

Before: manual assignment.

Now: a form or signup creates the HubSpot or Salesforce contact, the territory rule assigns the owner, and Slack or Teams notifies the rep with the answers and a follow-up task. Sign-off: none; the rep owns what happens next.

10. Lead scoring

Before: arrival-order lead work.

Now: title, company size, pricing page visits and trial activity from the CRM, product analytics and enrichment add or subtract points; crossing the threshold sets lifecycle stage and triggers routing. Sign-off: sales and marketing review the weights on a schedule.

11. Nurture sequences

Before: a weekly blast to everyone.

Now: list membership in the CRM (downloaded a guide, started a trial, went quiet) enrols contacts in an email sequence; a reply or purchase unenrols them. Sign-off: marketing approves the templates once.

12. Recurring reporting

Before: a manual Monday revenue sheet.

Now: on a schedule, Stripe, CRM and ad platform numbers are compared with the prior period and posted to Slack or Teams with what moved. Sign-off: none; the owner reads it.

Connect your CRM and billing in one click →

Client service and IT processes that route themselves

Support and IT service management run on queues, so each process sets queue, priority or owner.

13. Support ticket routing

Before: a manager sorted each ticket.

Now: channel, product area and plan set queue and priority in the helpdesk (Zendesk, Intercom or Pylon), with the CRM record linked. Sign-off: none; misroutes are reassigned.

14. Escalation rules

Before: breaches surfaced via complaints.

Now: SLA timer or keyword (outage, cancel, legal) raises priority, alerts the escalation channel and pages the on-call lead. Sign-off: the lead decides on refunds, credits or exceptions.

15. IT ticket intake and categorisation

Before: requests came by DM and email.

Now: a Slack or Teams form creates ticket; type sets category and assignee; standard requests (password reset, install from the approved list) follow documented steps. Sign-off: IT reclassifies when the category is wrong.

16. Access provisioning

Before: any admin granted access.

Now: an approved request grants role in the identity provider, seats follow groups, and the grant is logged with requester, approver and date. Sign-off: the app owner approves privileged roles.

Operations and document processes worth automating

17. Inventory threshold reorder

Before: stock was counted on Fridays and reordered by email.

Now: when on-hand stock in the ERP or inventory system falls below the reorder point, a purchase order at the standard quantity goes to the supplier portal or by email. In supply chain and inventory management, the order follows example 3’s approval route and its goods receipt feeds example 2’s three-way match. Sign-off: ops approves orders above a value cap.

18. Contract and document generation

Before: an account manager retyped CRM details into a template.

Now: in contract management, a closed-won deal fills the template from CRM fields, saves the document to the client folder in cloud storage and links it on the deal. Sign-off: legal reviews non-standard terms.

19. E-signature routing and filing

Before: the PDF was emailed, chased and filed anywhere.

Now: a generated document goes out for e-signatures in signer order, reminders follow the schedule, the signed copy is filed to storage and the CRM record is set to signed. Sign-off: the signer; filing needs none.

20. Cross-tool data sync

Before: one customer existed three times with three spellings.

Now: a record created or updated in the CRM pushes the agreed fields to billing, the helpdesk and the data warehouse; conflicts go to a review list, not overwritten. Sign-off: ops reviews the conflict list.

Industry-specific examples

The same numbered examples carry over to industry use cases, with a named person still signing off where the referenced example says so.

• Healthcare provider: credentialing and onboarding of clinical staff reuse examples 5 and 6. Patient intake forms follow example 15’s intake pattern, where a form creates the record and the request type sets the category and assignee.
• Insurer: claims processing intake, document check and routing work the way example 1 routes invoices. Extraction checks that the claim documents are present, and incomplete claims go to an exceptions queue for a person.
• Bank: every privileged access grant is logged with requester, approver and date, as in example 16. KYC document collection is routed and filed as in example 19.
• Construction firm: site material requests go through example 3’s approval route. Subcontractor document collection is routed and filed as in example 19.

One process built end to end: invoice approval

Examples 1 to 3 combined, at the step level:

1. Trigger: an email with a PDF attachment arrives in the AP mailbox.
2. Data source: vendor, invoice number, amount, due date and PO number are extracted and matched to the open POs.
3. Condition one: no PO number or unknown vendor. The invoice goes to the exceptions queue with the reason and stops there for a person.
4. Condition two: PO found and the total is within tolerance of the PO and the goods receipt. The bill is created and approved for payment.
5. Condition three: PO found but the total is above tolerance. The bill is created on hold, and an approval request goes to the budget owner in Slack or Teams with the variance shown.
6. Approval: approval releases the hold; rejection returns the invoice to AP with the comment attached.
7. Error path: extraction fails or the accounting system rejects the bill. The invoice is parked in a “needs attention” view and AP is notified once.
8. Notification: a daily summary of invoices paid, on hold and in exceptions is posted to the finance channel.

Setup detail is covered in the AI workflow automation guide.

Hand Viktor your invoice exceptions queue. Start free with up to $100 in credits, no credit card required →

Which process should you automate first?

Score each candidate on four factors.

• Volume: how many times a week it runs. A process that runs twice a month may not repay the setup.
• Rule clarity: can the decision be written as if-then without “it depends”?
• Error cost: a high error cost does not rule a process out; it moves the sign-off point closer to the action.
• Number of handoffs: each person the process passes through is a place it stalls.

Invoice processing, lead routing and leave requests score high on all four; sequencing is covered in 5 workflows to automate first.

BPA vs RPA vs BPM vs AI employees

The four terms overlap, from business process automation to business process management (BPM) and the AI employee; the split that matters is what the software acts on.

Where automated processes break and how to keep them running

A process that quietly stopped is worse than one never automated, because nobody is watching it. Why automations die quietly covers the pattern; each failure mode below has one mitigation.

• Silent errors: the run fails and nothing tells anyone. Mitigation: every run writes a status, and a weekly count compares runs against expected runs.
• Stale integrations: a token expires or a field is renamed and the process runs on with empty values. Mitigation: one named owner per integration and a test run after any schema change.
• Approval bottlenecks: the only approver is on leave and the queue grows. Mitigation: a backup approver and a time-out that escalates rather than auto-approves.
• Missing audit trails: a payment went out and nobody can say who approved it. Mitigation: the approval is recorded on the record itself, with the approver and the time.

Viktor applies the same approval step to every sensitive action, and everything is logged. His security page states that “Money moves, code pushes, and customer emails wait for your explicit approval”, and the approvals documentation shows that each card records who decided. Admins can require approval for individual integration tools.

Read the security posture →

See how the up to $100 in free credits works, no card required.

How Viktor runs these processes for your whole team

Viktor is an AI employee who lives in Slack and Microsoft Teams and connects to 9,800+ tools, most through one-click OAuth. For readers new to the idea, this is what an AI employee is: a hire who owns a job and executes it, not a tool that waits for rules.

Recurring processes run as scheduled tasks. Three of the examples above show how that looks:

• Recurring reporting (example 12): Viktor runs the Monday revenue, pipeline and ad spend recap as a scheduled task and posts it with what moved.
• Lead routing (example 9): Viktor starts when a form is submitted, then routes the lead to the right owner.
• Invoice processing (example 1): Viktor works the exceptions queue and waits for approval before anything is paid or sent externally.

He asks for approval before he sends external email or deploys to production. The request arrives as an approval card with Approve and Reject buttons, and a rejected action does not run.

Hire your first AI employee. Start free with up to $100 in credits, no credit card required →

FAQ

How do I automate a process step by step?

Name the trigger, list the systems the data passes through, write each decision as an if-then, and mark where a person approves. Build it against a small batch, compare the output with the manual result, then switch the trigger on.

How are RPA and BPA different?

BPA connects systems through their APIs and works on the data directly. RPA replays clicks and keystrokes through the screen of a system with no usable API, so it is the fallback for legacy systems, while BPA is the default where an API exists.

What are the top 5 RPA tools?

The five RPA platforms teams shortlist most often are below, with one alternative first for teams that would rather describe the outcome than script clicks.

1. Viktor: not screen RPA but an AI employee who works through APIs across 9,800+ tools and asks for approval before sensitive actions.
2. UiPath: offers “Unattended RPA” as an add-on, plus process mining on its Enterprise plan.
3. Microsoft Power Automate: lists “Attended and unattended RPA” next to its API connectors.
4. Automation Anywhere: a cloud RPA platform built around software bots for back-office processes.
5. Blue Prism: RPA aimed at regulated enterprises such as banks and insurers.
6. Tungsten Automation (formerly Kofax): RPA paired with document capture, common in invoice and claims processing.

How much time and money does each automation actually save?

The table below is a worked illustration, not a measured result. Baseline: a 25-person agency, a fully loaded cost of $60 per hour, an assumed $100 per month in tool cost for each process, and the manual minutes per run shown in the table. Hours after counts the exceptions and checks a person still handles.

Payback in weeks = (setup hours × hourly cost + monthly tool cost) ÷ (hours saved per week × hourly cost). Swap in your own volumes and minutes: time each process for a week before automating, then count exceptions and fixes for a week after. Low-volume processes such as recurring reporting pay back more slowly, which is why volume comes first in the scoring above.

Does every automated process need a human sign-off?

No: routing, filing and reporting examples on this page run without one. A sign-off belongs where the action is hard to reverse or costs money: paying a bill above tolerance, granting a privileged role, issuing a refund. Put it as close to that action as possible.