Free audit

Automation8 min read

Process automation: what to automate first in a small business

How to identify which tasks are worth automating, which tools to use, and how to work out whether it pays before spending a euro.

Automating is not buying software with “AI” on the front page. It is taking off your team’s plate the tasks a computer does better: copying data, notifying somebody, generating the same document every week.

The hard part is not technical. It is choosing well, because automating a bad process only makes it faster in the wrong direction.

The rule for deciding what to automate

A task is a good candidate if it meets all four:

  1. It repeats. At least several times a month.
  2. It has clear rules. If somebody always has to “take a look at it”, it is not ready.
  3. The data is already in a system. If it has to be transcribed from paper, digitalise the input first.
  4. Errors are detectable. If a fault could go unnoticed for months, you need a control before you automate.

If any of the four fails, fix that first. A process nobody has ever written down cannot be automated: it has to be documented first, and documenting it usually saves time on its own.

How to work out whether it pays

Before buying anything, run this calculation with your own numbers:

  • Hours per month the task consumes × hourly cost of whoever does it = gross monthly saving.
  • Implementation cost + first-year licences = investment.
  • Investment ÷ monthly saving = months to break even.

If it comes out under nine or ten months, go ahead. If it stretches to two years, there is probably something more profitable to do first.

Always add a margin: a company’s first automation takes longer than planned because it includes learning how to do it.

The five processes that come up most

1. Lead capture and follow-up

The website form sends an email, somebody reads it, notes it down and — sometimes — follows up. There is a good amount of revenue sitting between “sometimes” and “always”.

Automated: the form creates the record in the CRM, assigns an owner, sends an immediate reply to the customer, and creates a follow-up task after three days if nobody has responded.

Replying within the hour versus replying the next day changes the probability of closing substantially. That is the highest-value part, and it is purely mechanical.

2. Appointment reminders

Any business with a diary loses slots to forgetfulness. An automatic reminder by WhatsApp or SMS 24 hours ahead, with a reschedule option, recovers part of those slots without anybody phoning anybody.

3. Recurring invoicing

Issuing, sending, recording payment and flagging non-payment. If you bill monthly retainers, this is half a day a month that disappears.

4. Reports

If somebody spends hours copying data from three places into a monthly deck, that report can be a dashboard that updates itself. The time saved is real, but the more valuable part is that the data stops being a week out of date.

5. Customer or employee onboarding

Creating accounts, sending documentation, scheduling the first meeting, granting access. A checklist that fires by itself stops a step being forgotten exactly when first impressions matter.

Tools: start with what you already have

The sensible sequence is:

  1. What you already pay for. Your CRM, your invoicing software or your email client probably already have automations nobody has switched on. It is the cheapest option there is.
  2. Visual connectors (Zapier, Make and similar). They connect applications without code. Ideal to start with and for low-volume processes. They get expensive as volume grows.
  3. Custom development. When the process is central to your business, volume is high, or nothing off the shelf fits. More expensive up front, cheaper over a year.

A common mistake is jumping straight to step three. Another, just as common, is staying on step two once the connector already costs more per month than an amortised build.

Mistakes that cost

  • Automating without documenting. If the process only exists in one person’s head, the automation inherits their unwritten exceptions.
  • No manual override. Every automated process needs a way to intervene when the odd case shows up. And it always shows up.
  • Not telling the team. An automation nobody explained reads as surveillance or as a threat, and sabotages itself.
  • Never reviewing it. Automations break when an API or a form changes. Check they still run every quarter.

Where to start this week

Take your team’s most repetitive task and write it out step by step, exceptions included. That document alone will show you two or three things that can be simplified without automating anything.

Then automate the single most mechanical step. One. When it has run for a month without incident, move to the next.

If you want an outside view on which of your processes are worth it, tell us how you work and we will give you a straight answer.

Start with the audit. It is free and commits you to nothing.

We look at your site, your rankings and your competitors, and put in writing what we would change and in what order. Whether you use it with us or on your own.