Robotic process automation for small firms: worth it?

Key takeaways

Robotic process automation (RPA) is software that copies the clicks and keystrokes a person makes across screens, so it can run a fixed, high-volume task without changing the systems underneath. For small businesses it suits a narrow band of tasks: stable, rule-based, digital and frequent. Most small-business tasks fail one of those tests, which is where people still win.

• RPA copies screen actions; it does not understand them.
• It needs stable screens, clean inputs and fixed rules.
• Bots break when a system or form layout changes.
• Many SMEs get more from simple integrations plus a team.

What is robotic process automation?

Robotic process automation is software configured to follow the same on-screen steps a person would: open a system, copy a value, paste it elsewhere, click submit. It is useful where systems cannot be connected directly. It is not artificial intelligence by default; it follows the recorded steps exactly and stops when anything unexpected appears.

• How UK SMEs can use outsourcing to handle repetitive tasks
• AI versus BPO: what to automate and what to staff

Which tasks suit RPA?

RPA suits tasks that run hundreds of times a week, with structured digital inputs, fixed rules and screens that rarely change. Copying order data between two systems with no integration is a typical example. Tasks that involve reading unstructured emails, chasing people or handling frequent exceptions suit it poorly.

• Good fit: high volume, structured inputs, fixed rules.
• Good fit: two systems with no integration available.
• Poor fit: unstructured documents and free-text emails.
• Poor fit: frequent exceptions, phone calls or chasing.

• Double data entry: what rekeying costs and how to stop it
• Straight-through processing: cut the touches per task
• Workflow automation for small businesses: where to start

What does RPA take to run?

RPA needs someone to design the bot, test it, monitor it daily and fix it when a screen changes. In an illustrative example, a bot that saves 10 hours a week may need 2 to 3 hours a month of maintenance and a person to handle the items it rejects. These are example figures. Count the rejected items, not just the processed ones.

• Human in the loop: pairing AI tools with an outsourced team
• Exception handling: the few cases that drive most cost

When do people beat bots?

People beat bots when the task has messy inputs, frequent exceptions, low volume, systems that change often or a customer on the other end. In those cases a managed team working to a written method is quicker to set up and adapts without reprogramming. Often the best answer is a direct integration for the predictable step and a team for everything around it.

• Rule-based vs judgement tasks: what to hand over
• Data enrichment: fill the gaps in your customer records

What does this look like in practice?

A pattern we see: a small business buys an automation tool for a task that turns out to have a dozen exception types. The bot handles the simple items and the exceptions pile up. A team that handles exceptions daily rescues the project.

• See client work
• BPO Solutions for UK businesses

RPA readiness checklist

Check these before buying RPA.

• Hundreds of items a week.
• Structured digital inputs.
• Screens unchanged for six months.
• No direct integration available.
• A named person to monitor and fix it.
• A route for rejected items.

• The repetitive task fit score: ready to hand over?

Next step

Send us the task you are thinking of automating. We will say plainly whether it suits a bot, an integration, a managed team or a mix.