Insights to Help You Grow, Automate, and Lead

+27 83 383 7832

What Should You Automate First in Your Business?

What Should You Automate First in Your Business?

October 28, 2026•10 min read

5 Questions to Ask Before You Spend Money

Not every task needs AI. Not every process needs automation. And sometimes, the smartest decision is to fix what you're already doing before investing in another tool.

There's a good chance you've been told that automation can save your business hours every week.

Automate your emails. Automate your follow-ups. Automate your marketing. Automate your customer service.

Somewhere along the way, it starts to sound as though the ultimate goal is to have a business that runs entirely on its own.

But here's a question worth asking.

What happens when you automate something that wasn't working properly in the first place?

You might get faster results. Unfortunately, they might be the wrong results.

Before investing in another subscription or implementing another workflow, you need to know whether the work is actually worth automating.

And that starts with understanding your business, not shopping for technology.

Why automating a broken process rarely solves the problem

Imagine a consulting business receiving enquiries through its website, email and LinkedIn.

Sometimes the owner responds. Sometimes an assistant does. Occasionally, enquiries get overlooked because everyone assumes somebody else is handling them.

The business decides it needs automation.

A new system is introduced. Emails are acknowledged automatically, contacts are created in the CRM, and reminders are sent.

Everything looks more organised.

But enquiries are still being missed.

Why?

Because nobody established who was responsible for following up.

The automation was working. The underlying process wasn't.

This is one of the biggest mistakes businesses can make when implementing new technology.

They automate activities before defining responsibility, outcomes and exceptions.

The result can be an expensive system that performs an inefficient process more consistently.

Before automating anything, we need to answer a more important question.

What exactly should happen, and who is responsible for making sure it happens?

5 questions to ask before automating anything

These five questions provide a practical starting point for deciding where automation belongs in your business.

1. Does this task happen often enough to justify automation?

Not everything that can be automated should be.

Let's say you spend 15 minutes completing a task once every three months.

That's one hour a year.

If it takes several hours to configure and test an automation, and additional time to maintain it, have you really gained anything?

Now consider a different task.

Your business receives 20 enquiries a week. Each enquiry requires five minutes of administration.

That's 100 minutes every week.

Over four weeks, you're spending approximately six hours and forty minutes on that activity.

If a workflow could reduce the administrative time to two minutes per enquiry, you could potentially release four hours over the same period.

That's worth investigating.

Of course, released capacity isn't automatically a financial saving. The value depends on what you do with the time and what the automation costs to operate.

The question to ask: How frequently does the task happen, and how much time does it actually consume?

2. Can you explain exactly how the process should work?

This is where I'd encourage most business owners to start.

Before opening an automation platform, take a piece of paper and describe the process.

What starts it?

What information is needed?

What happens next?

Who is responsible?

What happens if something goes wrong?

For example, think about what happens when somebody books a consultation.

The business may need to confirm the appointment, update the CRM, provide preparation instructions and remind the client.

Those are predictable steps.

But what happens if the client cancels? What if they reschedule? What if they need urgent assistance?

A reliable system needs rules for those situations too.

If you can't explain the process clearly, it's probably not ready to be automated.

The question to ask: Could another person follow these instructions without repeatedly coming back to me for clarification?

3. Does the work require human judgment?

This is particularly important for professional services businesses.

A lawyer's expertise isn't simply the ability to send documents or schedule appointments.

An accountant's value isn't limited to collecting financial information.

A business coach isn't valuable because they can automatically send a motivational email.

Clients pay these professionals for their knowledge, experience, judgment and ability to understand individual circumstances.

Consider a law firm.

Automation could acknowledge a new enquiry, collect initial information through an appropriate secure process and notify the person responsible.

But determining whether the firm can accept a matter, considering potential conflicts or advising a client requires appropriate professional oversight.

Or consider a coaching business.

Automation can schedule a session, send preparation material and remind the client about the appointment.

But understanding why the client is struggling and deciding how best to support them is a different kind of work.

AI may help prepare information or identify questions worth considering. That doesn't make it responsible for the advice.

The question to ask: Would I be comfortable allowing this decision to happen without a qualified person reviewing it?

If the answer is no, design the workflow with an appropriate approval or human handoff.

4. What happens when the automation gets something wrong?

This question is often overlooked.

We tend to design workflows around what should happen when everything goes according to plan.

But real businesses don't operate entirely according to plan.

Someone submits incomplete information.

A client changes their mind.

An appointment is cancelled.

An urgent enquiry arrives outside business hours.

A connected application becomes unavailable.

What should happen?

If an automated process fails silently, you may not discover the problem until a customer complains.

That's why good automation needs more than a trigger and a series of actions.

It needs exception handling, alerts and someone responsible for checking the results.

Microsoft Power Automate, for example, supports approval workflows where an automated process pauses for a human decision before continuing.

You don't necessarily need that particular platform. The important principle is that the person remains involved at the right point.

The question to ask: How will we know if the automation fails, and who will fix the problem?

5. Will the result justify the full cost?

It's easy to compare monthly subscription prices.

But the advertised price doesn't tell you everything about the cost of automation.

You may also need to consider implementation, usage charges, maintenance, training, data security and time spent reviewing results.

Let's look at a hypothetical example.

Suppose a business introduces an automation that costs $25 per month in software fees.

It also requires $300 in implementation work, which the owner chooses to spread over the first 12 months for comparison.

That makes the estimated first-year cost:

Cost

Amount

Monthly software

$25

Annual software

$300

Initial implementation

$300

First-year total

$600

Monthly equivalent in year one

$50

That's before maintenance, human review, integration costs or usage charges.

Now suppose it releases two hours a month.

That's 24 hours of potential capacity across the year.

The business is spending $600 to release those 24 hours, equivalent to $25 per hour of capacity before other costs.

Would that be worthwhile?

Possibly.

If those hours allow the owner to deliver additional profitable work, it may be valuable.

If the automation simply creates more administration or moves the problem somewhere else, perhaps not.

The question to ask: What measurable improvement am I buying, and what will it cost to maintain?

A simple framework for deciding what to automate

Let's make this practical.

The following decision guide can help you determine whether a task needs automation, AI assistance or a person.

Type of work

Suggested approach

Repetitive, predictable tasks with clear rules

Consider a straightforward workflow

Tasks requiring information to be summarised or interpreted

Consider AI assistance with appropriate review

Sensitive decisions, professional advice and client commitments

Keep qualified humans responsible

Processes with unclear ownership or inconsistent steps

Document and improve the process first

Occasional tasks with limited benefit from automation

Consider continuing manually

This is a starting point, not a substitute for assessing privacy, professional obligations and operational risk.

The principle is simple.

Start with the least complicated solution that reliably gets the job done.

What this looks like in a professional services business

Let's use a law firm as an example.

There are several activities that happen before, during and after a client matter.

Some are ideal candidates for automation. Others need greater care.

Before the matter

A potential client submits an enquiry.

A system could record the information, send a suitable acknowledgement and notify the responsible team member.

Appointment confirmations and reminders could also be automated.

These activities support the relationship without replacing the professional's judgment.

During the matter

The firm may use systems to track deadlines, assign internal tasks and remind team members when updates are required.

But decisions involving legal advice, professional obligations and the client's individual circumstances remain under appropriate professional control.

After the matter

The firm may automate administrative steps such as sending an approved feedback request, recording completion information or reminding the team to close outstanding tasks.

Sensitive feedback and complaints should be routed to a person who can evaluate and respond appropriately.

Notice the pattern.

Much of the opportunity lies in the administrative work surrounding the professional service.

That's where systems can create capacity while allowing professionals to focus on the expertise their clients actually need.

The same principle applies to consultants, accountants and coaches.

Try this seven-day automation audit

You don't need to purchase software to begin improving your business.

For the next seven working days, record recurring tasks that consume your time.

Use a simple table.

Task

How often?

Time required

Human judgment?

Next decision

Capture new enquiries

Daily

5 minutes

Limited

Document and test

Prepare complex client advice

Variable

60 minutes

High

Human-led

Send appointment reminders

Daily

3 minutes

Low

Check existing CRM

Respond to complaints

Occasionally

Variable

High

Human-led

These are illustrative entries. Your own figures will be different.

At the end of the week, look for tasks that are:

  • Frequent enough to matter.

  • Predictable and easy to describe.

  • Relatively low risk.

  • Time-consuming without adding much direct client value.

  • Possible to test without disrupting the business.

Choose just one.

Document the process. Identify the person responsible. Decide what a successful outcome looks like.

Then see whether your existing software can automate it.

If you test an enquiry workflow, measure whether enquiries are captured correctly, assigned to the right person and followed up within your agreed timeframe.

You can also track how many exceptions require human intervention.

Start with a small supervised test before expanding the automation.

The real goal isn't to automate everything

It's to create a business that works better.

Sometimes that means introducing AI.

Sometimes it means building a straightforward workflow.

Sometimes it means removing an unnecessary step, improving communication or making somebody responsible for a process that previously had no clear owner.

And sometimes the best answer is to leave a task exactly where it belongs, with a person.

Business owners don't need more complexity.

They need systems that give them better control, reduce unnecessary work and support the relationships that make their businesses valuable.

Before asking what you can automate, ask what actually needs improving.

Your expertise delivers the service. Your systems should support everything around it.


Where is your business losing time?

Before investing in another tool, it may be worth discovering where unnecessary work, repeated administration and unclear processes are already costing you time.

Our BizWithAI Process Waste Assessment helps you take a closer look at those areas and identify where improvements may be possible.

Take the Process Waste Assessment

BizWithAI helps professional services businesses build better systems around the work they already do. Before service. After service. And every handoff in between.

Tilly Davies

Tilly Davies

Tech Smart Marketer - a visionary with over 40 years of unparalleled experience in B2B Business Analysis, IT, Finance and digital marketing.

LinkedIn logo icon
Instagram logo icon
Youtube logo icon
Back to Blog
Comments
Be the first to comment
Comments
Be the first to comment